diff --git "a/checkpoint-14000/trainer_state.json" "b/checkpoint-14000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-14000/trainer_state.json"
@@ -0,0 +1,16879 @@
+{
+  "best_metric": 0.02988557517528534,
+  "best_model_checkpoint": "wav2vec2-base-pemlsb-la/checkpoint-14000",
+  "epoch": 1.553955437165293,
+  "global_step": 14000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.08,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9994446912483344e-05,
+      "loss": 0.0956,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9988893824966684e-05,
+      "loss": 0.1378,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9983340737450023e-05,
+      "loss": 0.0434,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9977787649933366e-05,
+      "loss": 0.1188,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9972234562416705e-05,
+      "loss": 0.1167,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9966681474900048e-05,
+      "loss": 0.1105,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9961128387383387e-05,
+      "loss": 0.1941,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9955575299866727e-05,
+      "loss": 0.2953,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995002221235007e-05,
+      "loss": 0.1396,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.994446912483341e-05,
+      "loss": 0.0832,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9938916037316748e-05,
+      "loss": 0.1062,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993336294980009e-05,
+      "loss": 0.1134,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992780986228343e-05,
+      "loss": 0.0965,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9922256774766773e-05,
+      "loss": 0.0549,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9916703687250112e-05,
+      "loss": 0.0813,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9911150599733452e-05,
+      "loss": 0.1101,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9905597512216795e-05,
+      "loss": 0.202,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9900044424700134e-05,
+      "loss": 0.367,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9894491337183473e-05,
+      "loss": 0.1122,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9888938249666816e-05,
+      "loss": 0.1137,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.988338516215016e-05,
+      "loss": 0.0945,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9877832074633498e-05,
+      "loss": 0.1107,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9872278987116838e-05,
+      "loss": 0.0774,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.986672589960018e-05,
+      "loss": 0.0625,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.986117281208352e-05,
+      "loss": 0.0663,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9855619724566862e-05,
+      "loss": 0.1929,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9850066637050202e-05,
+      "loss": 0.1206,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.984451354953354e-05,
+      "loss": 0.433,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9838960462016884e-05,
+      "loss": 0.103,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9833407374500223e-05,
+      "loss": 0.0935,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9827854286983566e-05,
+      "loss": 0.1095,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9822301199466905e-05,
+      "loss": 0.1218,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9816748111950245e-05,
+      "loss": 0.0888,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9811195024433588e-05,
+      "loss": 0.0929,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9805641936916927e-05,
+      "loss": 0.1013,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9800088849400266e-05,
+      "loss": 0.136,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.979453576188361e-05,
+      "loss": 0.1258,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.978898267436695e-05,
+      "loss": 0.3627,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.978342958685029e-05,
+      "loss": 0.0882,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.977787649933363e-05,
+      "loss": 0.1098,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.977232341181697e-05,
+      "loss": 0.0994,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9766770324300313e-05,
+      "loss": 0.1581,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9761217236783655e-05,
+      "loss": 0.0698,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9755664149266995e-05,
+      "loss": 0.1023,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9750111061750334e-05,
+      "loss": 0.0995,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9744557974233677e-05,
+      "loss": 0.1318,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9739004886717016e-05,
+      "loss": 0.2402,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9733451799200356e-05,
+      "loss": 0.4634,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.97278987116837e-05,
+      "loss": 0.0771,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9722345624167038e-05,
+      "loss": 0.0877,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.971679253665038e-05,
+      "loss": 0.1057,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.971123944913372e-05,
+      "loss": 0.094,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.970568636161706e-05,
+      "loss": 0.0814,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9700133274100402e-05,
+      "loss": 0.0621,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.969458018658374e-05,
+      "loss": 0.111,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9689027099067084e-05,
+      "loss": 0.1037,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9683474011550424e-05,
+      "loss": 0.1733,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9677920924033763e-05,
+      "loss": 0.3142,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9672367836517106e-05,
+      "loss": 0.0859,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.966681474900045e-05,
+      "loss": 0.1049,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9661261661483784e-05,
+      "loss": 0.1117,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9655708573967127e-05,
+      "loss": 0.1147,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.965015548645047e-05,
+      "loss": 0.0874,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.964460239893381e-05,
+      "loss": 0.1098,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.963904931141715e-05,
+      "loss": 0.085,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.963349622390049e-05,
+      "loss": 0.1585,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962794313638383e-05,
+      "loss": 0.2213,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9622390048867174e-05,
+      "loss": 0.4533,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9616836961350513e-05,
+      "loss": 0.1004,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9611283873833852e-05,
+      "loss": 0.0968,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9605730786317195e-05,
+      "loss": 0.1071,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9600177698800534e-05,
+      "loss": 0.1426,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9594624611283874e-05,
+      "loss": 0.1237,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9589071523767217e-05,
+      "loss": 0.0596,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9583518436250556e-05,
+      "loss": 0.0578,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.95779653487339e-05,
+      "loss": 0.1257,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9572412261217238e-05,
+      "loss": 0.1492,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9566859173700577e-05,
+      "loss": 0.2997,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.956130608618392e-05,
+      "loss": 0.1145,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.955575299866726e-05,
+      "loss": 0.1076,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.95501999111506e-05,
+      "loss": 0.0894,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.954464682363394e-05,
+      "loss": 0.1168,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9539093736117284e-05,
+      "loss": 0.0885,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9533540648600624e-05,
+      "loss": 0.0653,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9527987561083967e-05,
+      "loss": 0.0617,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9522434473567306e-05,
+      "loss": 0.1256,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9516881386050645e-05,
+      "loss": 0.238,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9511328298533988e-05,
+      "loss": 0.326,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9505775211017327e-05,
+      "loss": 0.118,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9500222123500667e-05,
+      "loss": 0.0939,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.949466903598401e-05,
+      "loss": 0.121,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.948911594846735e-05,
+      "loss": 0.1211,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.948356286095069e-05,
+      "loss": 0.0962,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.947800977343403e-05,
+      "loss": 0.059,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.947245668591737e-05,
+      "loss": 0.1039,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9468014215904045e-05,
+      "loss": 0.1474,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9462461128387385e-05,
+      "loss": 0.1206,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9456908040870724e-05,
+      "loss": 0.4527,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9451354953354067e-05,
+      "loss": 0.1067,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944580186583741e-05,
+      "loss": 0.094,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944024877832075e-05,
+      "loss": 0.0877,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9434695690804088e-05,
+      "loss": 0.0908,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.942914260328743e-05,
+      "loss": 0.0999,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.942358951577077e-05,
+      "loss": 0.0366,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.941803642825411e-05,
+      "loss": 0.0938,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9412483340737452e-05,
+      "loss": 0.1235,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9406930253220792e-05,
+      "loss": 0.1556,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9401377165704135e-05,
+      "loss": 0.2944,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9395824078187474e-05,
+      "loss": 0.0929,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9390270990670813e-05,
+      "loss": 0.1046,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9384717903154156e-05,
+      "loss": 0.0832,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9379164815637495e-05,
+      "loss": 0.105,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9373611728120835e-05,
+      "loss": 0.1068,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9368058640604178e-05,
+      "loss": 0.0657,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9362505553087517e-05,
+      "loss": 0.1071,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.935695246557086e-05,
+      "loss": 0.132,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.93513993780542e-05,
+      "loss": 0.1616,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.934584629053754e-05,
+      "loss": 0.389,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.934029320302088e-05,
+      "loss": 0.0847,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9334740115504224e-05,
+      "loss": 0.0961,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9329187027987563e-05,
+      "loss": 0.0977,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9323633940470903e-05,
+      "loss": 0.1434,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9318080852954245e-05,
+      "loss": 0.0628,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9312527765437585e-05,
+      "loss": 0.1,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9306974677920928e-05,
+      "loss": 0.1099,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9301421590404267e-05,
+      "loss": 0.0929,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9295868502887606e-05,
+      "loss": 0.1874,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.929031541537095e-05,
+      "loss": 0.2716,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.928476232785429e-05,
+      "loss": 0.0824,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9279209240337628e-05,
+      "loss": 0.1086,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.927365615282097e-05,
+      "loss": 0.0988,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.926810306530431e-05,
+      "loss": 0.0907,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9262549977787653e-05,
+      "loss": 0.0858,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9256996890270992e-05,
+      "loss": 0.0548,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.925144380275433e-05,
+      "loss": 0.1034,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9245890715237674e-05,
+      "loss": 0.0921,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9240337627721014e-05,
+      "loss": 0.1742,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9234784540204353e-05,
+      "loss": 0.3367,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9229231452687696e-05,
+      "loss": 0.0953,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.922367836517104e-05,
+      "loss": 0.078,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9218125277654378e-05,
+      "loss": 0.1181,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9212572190137717e-05,
+      "loss": 0.0898,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.920701910262106e-05,
+      "loss": 0.0577,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.92014660151044e-05,
+      "loss": 0.0555,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9195912927587742e-05,
+      "loss": 0.1174,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.919035984007108e-05,
+      "loss": 0.1356,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.918480675255442e-05,
+      "loss": 0.1978,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9179253665037764e-05,
+      "loss": 0.3164,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9173700577521103e-05,
+      "loss": 0.0961,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9168147490004446e-05,
+      "loss": 0.0904,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9162594402487785e-05,
+      "loss": 0.1158,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9157041314971124e-05,
+      "loss": 0.1316,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9151488227454467e-05,
+      "loss": 0.076,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9145935139937807e-05,
+      "loss": 0.0589,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9140382052421146e-05,
+      "loss": 0.0889,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.913482896490449e-05,
+      "loss": 0.1223,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9129275877387828e-05,
+      "loss": 0.1641,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.912372278987117e-05,
+      "loss": 0.3307,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.911816970235451e-05,
+      "loss": 0.1,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.911261661483785e-05,
+      "loss": 0.0979,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9107063527321192e-05,
+      "loss": 0.1066,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9101510439804535e-05,
+      "loss": 0.1103,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9095957352287874e-05,
+      "loss": 0.0711,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9090404264771214e-05,
+      "loss": 0.0922,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9084851177254557e-05,
+      "loss": 0.0949,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9079298089737896e-05,
+      "loss": 0.0693,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9073745002221235e-05,
+      "loss": 0.1918,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9068191914704578e-05,
+      "loss": 0.3019,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9062638827187917e-05,
+      "loss": 0.093,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.905708573967126e-05,
+      "loss": 0.1074,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.90515326521546e-05,
+      "loss": 0.1087,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.904597956463794e-05,
+      "loss": 0.1158,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.904042647712128e-05,
+      "loss": 0.0637,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.903487338960462e-05,
+      "loss": 0.0768,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9029320302087964e-05,
+      "loss": 0.0905,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9023767214571303e-05,
+      "loss": 0.0926,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9018214127054642e-05,
+      "loss": 0.1894,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9012661039537985e-05,
+      "loss": 0.412,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9007107952021328e-05,
+      "loss": 0.0885,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9001554864504664e-05,
+      "loss": 0.1324,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8996001776988007e-05,
+      "loss": 0.1101,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.899044868947135e-05,
+      "loss": 0.1102,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.898489560195469e-05,
+      "loss": 0.0965,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8979342514438028e-05,
+      "loss": 0.0451,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.897378942692137e-05,
+      "loss": 0.0892,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.896823633940471e-05,
+      "loss": 0.1309,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8962683251888053e-05,
+      "loss": 0.2524,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8957130164371392e-05,
+      "loss": 0.3986,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8951577076854732e-05,
+      "loss": 0.0876,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8946023989338075e-05,
+      "loss": 0.1454,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8940470901821414e-05,
+      "loss": 0.0976,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8934917814304753e-05,
+      "loss": 0.1096,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8929364726788096e-05,
+      "loss": 0.0861,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8923811639271435e-05,
+      "loss": 0.0918,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8918258551754778e-05,
+      "loss": 0.0726,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8912705464238118e-05,
+      "loss": 0.1302,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8907152376721457e-05,
+      "loss": 0.1552,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.89015992892048e-05,
+      "loss": 0.2767,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.889604620168814e-05,
+      "loss": 0.0916,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.889049311417148e-05,
+      "loss": 0.1051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.888494002665482e-05,
+      "loss": 0.0975,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8879386939138164e-05,
+      "loss": 0.1404,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8873833851621503e-05,
+      "loss": 0.0996,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8868280764104846e-05,
+      "loss": 0.0673,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8862727676588185e-05,
+      "loss": 0.0836,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8857174589071525e-05,
+      "loss": 0.1149,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8851621501554868e-05,
+      "loss": 0.2039,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8846068414038207e-05,
+      "loss": 0.3666,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8840515326521546e-05,
+      "loss": 0.1245,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.883496223900489e-05,
+      "loss": 0.1168,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.882940915148823e-05,
+      "loss": 0.1029,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.882385606397157e-05,
+      "loss": 0.1302,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.881830297645491e-05,
+      "loss": 0.0962,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.881274988893825e-05,
+      "loss": 0.0983,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8807196801421593e-05,
+      "loss": 0.1008,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8801643713904932e-05,
+      "loss": 0.1457,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.879609062638827e-05,
+      "loss": 0.1688,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8790537538871614e-05,
+      "loss": 0.4036,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8784984451354954e-05,
+      "loss": 0.0819,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8779431363838296e-05,
+      "loss": 0.113,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8773878276321636e-05,
+      "loss": 0.1456,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8768325188804975e-05,
+      "loss": 0.1137,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8762772101288318e-05,
+      "loss": 0.1032,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.875721901377166e-05,
+      "loss": 0.0595,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8751665926255e-05,
+      "loss": 0.0963,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.874611283873834e-05,
+      "loss": 0.093,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8740559751221682e-05,
+      "loss": 0.1282,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.873500666370502e-05,
+      "loss": 0.3287,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8729453576188364e-05,
+      "loss": 0.109,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8723900488671704e-05,
+      "loss": 0.093,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8718347401155043e-05,
+      "loss": 0.1131,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8712794313638386e-05,
+      "loss": 0.1085,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8707241226121725e-05,
+      "loss": 0.1316,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8701688138605064e-05,
+      "loss": 0.129,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8696135051088407e-05,
+      "loss": 0.0762,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8690581963571747e-05,
+      "loss": 0.1121,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.868502887605509e-05,
+      "loss": 0.1596,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.867947578853843e-05,
+      "loss": 0.3276,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8673922701021768e-05,
+      "loss": 0.1237,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.866836961350511e-05,
+      "loss": 0.0868,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8662816525988454e-05,
+      "loss": 0.1102,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.865726343847179e-05,
+      "loss": 0.1202,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8651710350955132e-05,
+      "loss": 0.0971,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8646157263438475e-05,
+      "loss": 0.0634,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8640604175921814e-05,
+      "loss": 0.1133,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8635051088405154e-05,
+      "loss": 0.1406,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8629498000888497e-05,
+      "loss": 0.1389,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8623944913371836e-05,
+      "loss": 0.3533,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.861839182585518e-05,
+      "loss": 0.1312,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8612838738338518e-05,
+      "loss": 0.0902,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8607285650821857e-05,
+      "loss": 0.1013,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.86017325633052e-05,
+      "loss": 0.1126,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.859617947578854e-05,
+      "loss": 0.1037,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.859062638827188e-05,
+      "loss": 0.0653,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.858507330075522e-05,
+      "loss": 0.0894,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.857952021323856e-05,
+      "loss": 0.0891,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8573967125721904e-05,
+      "loss": 0.1735,
+      "step": 1295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8568414038205243e-05,
+      "loss": 0.3859,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8562860950688583e-05,
+      "loss": 0.0907,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8557307863171925e-05,
+      "loss": 0.0897,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8551754775655265e-05,
+      "loss": 0.0812,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8546201688138607e-05,
+      "loss": 0.1427,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8540648600621947e-05,
+      "loss": 0.0853,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.853509551310529e-05,
+      "loss": 0.0821,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.852954242558863e-05,
+      "loss": 0.1077,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.852398933807197e-05,
+      "loss": 0.1554,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.851843625055531e-05,
+      "loss": 0.1472,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.851288316303865e-05,
+      "loss": 0.3838,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8507330075521993e-05,
+      "loss": 0.1012,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8501776988005332e-05,
+      "loss": 0.0975,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8496223900488672e-05,
+      "loss": 0.1068,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8490670812972015e-05,
+      "loss": 0.0981,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8485117725455354e-05,
+      "loss": 0.109,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8479564637938697e-05,
+      "loss": 0.1133,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8474011550422036e-05,
+      "loss": 0.0995,
+      "step": 1385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8468458462905375e-05,
+      "loss": 0.1472,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8462905375388718e-05,
+      "loss": 0.1471,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8457352287872058e-05,
+      "loss": 0.3345,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8451799200355397e-05,
+      "loss": 0.0969,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.844624611283874e-05,
+      "loss": 0.0834,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.844069302532208e-05,
+      "loss": 0.0904,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8435139937805422e-05,
+      "loss": 0.1053,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8429586850288765e-05,
+      "loss": 0.128,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.84240337627721e-05,
+      "loss": 0.0652,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8418480675255443e-05,
+      "loss": 0.0541,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8412927587738786e-05,
+      "loss": 0.1123,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8407374500222125e-05,
+      "loss": 0.1649,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8401821412705465e-05,
+      "loss": 0.3193,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8396268325188808e-05,
+      "loss": 0.1042,
+      "step": 1455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8390715237672147e-05,
+      "loss": 0.105,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.838516215015549e-05,
+      "loss": 0.1443,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.837960906263883e-05,
+      "loss": 0.128,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.837405597512217e-05,
+      "loss": 0.0748,
+      "step": 1475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.836850288760551e-05,
+      "loss": 0.0674,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.836294980008885e-05,
+      "loss": 0.1105,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.835739671257219e-05,
+      "loss": 0.1129,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8351843625055533e-05,
+      "loss": 0.183,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8346290537538872e-05,
+      "loss": 0.4749,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8340737450022215e-05,
+      "loss": 0.0772,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8335184362505554e-05,
+      "loss": 0.1174,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8329631274988894e-05,
+      "loss": 0.1242,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8324078187472236e-05,
+      "loss": 0.1117,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.831852509995558e-05,
+      "loss": 0.0696,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8312972012438915e-05,
+      "loss": 0.0599,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8307418924922258e-05,
+      "loss": 0.0583,
+      "step": 1535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.83018658374056e-05,
+      "loss": 0.1187,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.829631274988894e-05,
+      "loss": 0.197,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.829075966237228e-05,
+      "loss": 0.291,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8285206574855622e-05,
+      "loss": 0.0863,
+      "step": 1555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.827965348733896e-05,
+      "loss": 0.1142,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8274100399822304e-05,
+      "loss": 0.099,
+      "step": 1565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8268547312305644e-05,
+      "loss": 0.1095,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8262994224788983e-05,
+      "loss": 0.0762,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8257441137272326e-05,
+      "loss": 0.0783,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8251888049755665e-05,
+      "loss": 0.0961,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8246334962239008e-05,
+      "loss": 0.1519,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8240781874722347e-05,
+      "loss": 0.2982,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8235228787205687e-05,
+      "loss": 0.3164,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.822967569968903e-05,
+      "loss": 0.0642,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.822412261217237e-05,
+      "loss": 0.0919,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8218569524655708e-05,
+      "loss": 0.1066,
+      "step": 1615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.821301643713905e-05,
+      "loss": 0.1179,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.820746334962239e-05,
+      "loss": 0.0849,
+      "step": 1625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8201910262105733e-05,
+      "loss": 0.0702,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8196357174589072e-05,
+      "loss": 0.1024,
+      "step": 1635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8190804087072415e-05,
+      "loss": 0.1288,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8185250999555754e-05,
+      "loss": 0.2155,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8179697912039097e-05,
+      "loss": 0.3855,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8174144824522437e-05,
+      "loss": 0.0816,
+      "step": 1655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8168591737005776e-05,
+      "loss": 0.0981,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.816303864948912e-05,
+      "loss": 0.1047,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8157485561972458e-05,
+      "loss": 0.1031,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8151932474455797e-05,
+      "loss": 0.081,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.814637938693914e-05,
+      "loss": 0.0726,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.814082629942248e-05,
+      "loss": 0.097,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8135273211905822e-05,
+      "loss": 0.1084,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.812972012438916e-05,
+      "loss": 0.1585,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.81241670368725e-05,
+      "loss": 0.3579,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8118613949355844e-05,
+      "loss": 0.0689,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8113060861839183e-05,
+      "loss": 0.0951,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8107507774322526e-05,
+      "loss": 0.0974,
+      "step": 1715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8101954686805865e-05,
+      "loss": 0.1083,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8096401599289205e-05,
+      "loss": 0.0747,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8090848511772547e-05,
+      "loss": 0.074,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.808529542425589e-05,
+      "loss": 0.0905,
+      "step": 1735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8079742336739226e-05,
+      "loss": 0.1717,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.807418924922257e-05,
+      "loss": 0.1754,
+      "step": 1745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.806863616170591e-05,
+      "loss": 0.4073,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.806308307418925e-05,
+      "loss": 0.0598,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805752998667259e-05,
+      "loss": 0.0926,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8051976899155933e-05,
+      "loss": 0.0822,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8046423811639272e-05,
+      "loss": 0.1055,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8040870724122615e-05,
+      "loss": 0.1027,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8035317636605955e-05,
+      "loss": 0.0651,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8029764549089294e-05,
+      "loss": 0.0937,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8024211461572637e-05,
+      "loss": 0.1496,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8018658374055976e-05,
+      "loss": 0.1447,
+      "step": 1795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8014215904042648e-05,
+      "loss": 0.5813,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800866281652599e-05,
+      "loss": 0.121,
+      "step": 1805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8003109729009333e-05,
+      "loss": 0.1069,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.799755664149267e-05,
+      "loss": 0.0923,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7992003553976012e-05,
+      "loss": 0.12,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7986450466459355e-05,
+      "loss": 0.0846,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7980897378942694e-05,
+      "loss": 0.1039,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7975344291426033e-05,
+      "loss": 0.1386,
+      "step": 1835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7969791203909376e-05,
+      "loss": 0.1327,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7964238116392715e-05,
+      "loss": 0.1693,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7958685028876058e-05,
+      "loss": 0.4163,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7953131941359398e-05,
+      "loss": 0.0809,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7947578853842737e-05,
+      "loss": 0.1079,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794202576632608e-05,
+      "loss": 0.133,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793647267880942e-05,
+      "loss": 0.1613,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793091959129276e-05,
+      "loss": 0.0594,
+      "step": 1875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79253665037761e-05,
+      "loss": 0.0836,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791981341625944e-05,
+      "loss": 0.1182,
+      "step": 1885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7914260328742783e-05,
+      "loss": 0.0962,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7908707241226123e-05,
+      "loss": 0.1491,
+      "step": 1895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7903154153709462e-05,
+      "loss": 0.4326,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7897601066192805e-05,
+      "loss": 0.0958,
+      "step": 1905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7892047978676144e-05,
+      "loss": 0.1066,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7886494891159487e-05,
+      "loss": 0.1029,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7880941803642826e-05,
+      "loss": 0.0875,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787538871612617e-05,
+      "loss": 0.0837,
+      "step": 1925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786983562860951e-05,
+      "loss": 0.0565,
+      "step": 1930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786428254109285e-05,
+      "loss": 0.1131,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785872945357619e-05,
+      "loss": 0.1341,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785317636605953e-05,
+      "loss": 0.1491,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7847623278542873e-05,
+      "loss": 0.3896,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7842070191026212e-05,
+      "loss": 0.075,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783651710350955e-05,
+      "loss": 0.1017,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7830964015992894e-05,
+      "loss": 0.1047,
+      "step": 1965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7825410928476234e-05,
+      "loss": 0.0862,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7819857840959576e-05,
+      "loss": 0.1087,
+      "step": 1975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7814304753442916e-05,
+      "loss": 0.0662,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7808751665926255e-05,
+      "loss": 0.0751,
+      "step": 1985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7803198578409598e-05,
+      "loss": 0.1606,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7797645490892937e-05,
+      "loss": 0.184,
+      "step": 1995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7792092403376277e-05,
+      "loss": 0.3407,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.050165969878435135,
+      "eval_runtime": 1712.7733,
+      "eval_samples_per_second": 9.352,
+      "eval_steps_per_second": 2.338,
+      "eval_wer": 0.13122971510998918,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778653931585962e-05,
+      "loss": 0.1099,
+      "step": 2005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778098622834296e-05,
+      "loss": 0.1189,
+      "step": 2010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77754331408263e-05,
+      "loss": 0.0952,
+      "step": 2015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7769880053309644e-05,
+      "loss": 0.0993,
+      "step": 2020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776432696579298e-05,
+      "loss": 0.1227,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7758773878276323e-05,
+      "loss": 0.0403,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7753220790759666e-05,
+      "loss": 0.0698,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7747667703243005e-05,
+      "loss": 0.1324,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7742114615726344e-05,
+      "loss": 0.1785,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7736561528209687e-05,
+      "loss": 0.2467,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7731008440693027e-05,
+      "loss": 0.0919,
+      "step": 2055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772545535317637e-05,
+      "loss": 0.0911,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.771990226565971e-05,
+      "loss": 0.098,
+      "step": 2065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7714349178143048e-05,
+      "loss": 0.1045,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770879609062639e-05,
+      "loss": 0.09,
+      "step": 2075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770324300310973e-05,
+      "loss": 0.0827,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769768991559307e-05,
+      "loss": 0.072,
+      "step": 2085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7692136828076412e-05,
+      "loss": 0.0998,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768658374055975e-05,
+      "loss": 0.205,
+      "step": 2095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7681030653043094e-05,
+      "loss": 0.3571,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7675477565526434e-05,
+      "loss": 0.122,
+      "step": 2105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7669924478009773e-05,
+      "loss": 0.1288,
+      "step": 2110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7664371390493116e-05,
+      "loss": 0.1193,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.765881830297646e-05,
+      "loss": 0.1113,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7653265215459795e-05,
+      "loss": 0.102,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7647712127943137e-05,
+      "loss": 0.038,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.764215904042648e-05,
+      "loss": 0.0566,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.763660595290982e-05,
+      "loss": 0.0881,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.763105286539316e-05,
+      "loss": 0.1674,
+      "step": 2145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76254997778765e-05,
+      "loss": 0.2494,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.761994669035984e-05,
+      "loss": 0.0807,
+      "step": 2155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7614393602843184e-05,
+      "loss": 0.0811,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7608840515326523e-05,
+      "loss": 0.117,
+      "step": 2165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7603287427809863e-05,
+      "loss": 0.1444,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7597734340293205e-05,
+      "loss": 0.0817,
+      "step": 2175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7592181252776545e-05,
+      "loss": 0.0418,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7586628165259887e-05,
+      "loss": 0.1151,
+      "step": 2185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7581075077743227e-05,
+      "loss": 0.0961,
+      "step": 2190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7575521990226566e-05,
+      "loss": 0.1609,
+      "step": 2195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.756996890270991e-05,
+      "loss": 0.3116,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7564415815193248e-05,
+      "loss": 0.0887,
+      "step": 2205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558862727676588e-05,
+      "loss": 0.087,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755330964015993e-05,
+      "loss": 0.0946,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754775655264327e-05,
+      "loss": 0.1125,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542203465126612e-05,
+      "loss": 0.0796,
+      "step": 2225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536650377609952e-05,
+      "loss": 0.0459,
+      "step": 2230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531097290093295e-05,
+      "loss": 0.0971,
+      "step": 2235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525544202576634e-05,
+      "loss": 0.1492,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519991115059977e-05,
+      "loss": 0.171,
+      "step": 2245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514438027543316e-05,
+      "loss": 0.3832,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508884940026655e-05,
+      "loss": 0.0961,
+      "step": 2255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503331852509998e-05,
+      "loss": 0.0928,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497778764993338e-05,
+      "loss": 0.1037,
+      "step": 2265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492225677476677e-05,
+      "loss": 0.1006,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748667258996002e-05,
+      "loss": 0.1022,
+      "step": 2275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748111950244336e-05,
+      "loss": 0.0412,
+      "step": 2280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7475566414926702e-05,
+      "loss": 0.1383,
+      "step": 2285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747001332741004e-05,
+      "loss": 0.1163,
+      "step": 2290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746446023989338e-05,
+      "loss": 0.1866,
+      "step": 2295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458907152376723e-05,
+      "loss": 0.2643,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453354064860063e-05,
+      "loss": 0.0854,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447800977343405e-05,
+      "loss": 0.0796,
+      "step": 2310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442247889826745e-05,
+      "loss": 0.1129,
+      "step": 2315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436694802310084e-05,
+      "loss": 0.0971,
+      "step": 2320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431141714793427e-05,
+      "loss": 0.0693,
+      "step": 2325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742558862727677e-05,
+      "loss": 0.0716,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420035539760106e-05,
+      "loss": 0.1012,
+      "step": 2335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741448245224345e-05,
+      "loss": 0.1319,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740892936472679e-05,
+      "loss": 0.2252,
+      "step": 2345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740337627721013e-05,
+      "loss": 0.2975,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739782318969347e-05,
+      "loss": 0.0746,
+      "step": 2355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392270102176813e-05,
+      "loss": 0.0867,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386717014660152e-05,
+      "loss": 0.1057,
+      "step": 2365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381163927143495e-05,
+      "loss": 0.1079,
+      "step": 2370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375610839626834e-05,
+      "loss": 0.0656,
+      "step": 2375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370057752110174e-05,
+      "loss": 0.077,
+      "step": 2380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364504664593516e-05,
+      "loss": 0.1688,
+      "step": 2385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7358951577076856e-05,
+      "loss": 0.1224,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7353398489560195e-05,
+      "loss": 0.1352,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7347845402043538e-05,
+      "loss": 0.2779,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7342292314526877e-05,
+      "loss": 0.0789,
+      "step": 2405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.733673922701022e-05,
+      "loss": 0.0798,
+      "step": 2410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.733118613949356e-05,
+      "loss": 0.088,
+      "step": 2415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.73256330519769e-05,
+      "loss": 0.0995,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.732007996446024e-05,
+      "loss": 0.1002,
+      "step": 2425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7314526876943584e-05,
+      "loss": 0.0484,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7308973789426924e-05,
+      "loss": 0.0947,
+      "step": 2435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7303420701910263e-05,
+      "loss": 0.121,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7297867614393606e-05,
+      "loss": 0.2656,
+      "step": 2445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7292314526876945e-05,
+      "loss": 0.3153,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7286761439360288e-05,
+      "loss": 0.1041,
+      "step": 2455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7281208351843627e-05,
+      "loss": 0.1142,
+      "step": 2460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7275655264326967e-05,
+      "loss": 0.0924,
+      "step": 2465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.727010217681031e-05,
+      "loss": 0.1132,
+      "step": 2470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.726454908929365e-05,
+      "loss": 0.1095,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7258996001776988e-05,
+      "loss": 0.0713,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.725344291426033e-05,
+      "loss": 0.1302,
+      "step": 2485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.724788982674367e-05,
+      "loss": 0.1497,
+      "step": 2490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7242336739227013e-05,
+      "loss": 0.1711,
+      "step": 2495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7236783651710352e-05,
+      "loss": 0.3014,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.723123056419369e-05,
+      "loss": 0.086,
+      "step": 2505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7225677476677034e-05,
+      "loss": 0.0806,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7220124389160374e-05,
+      "loss": 0.0961,
+      "step": 2515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7214571301643713e-05,
+      "loss": 0.1323,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7209018214127056e-05,
+      "loss": 0.0683,
+      "step": 2525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7203465126610395e-05,
+      "loss": 0.0736,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7197912039093738e-05,
+      "loss": 0.0852,
+      "step": 2535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7192358951577077e-05,
+      "loss": 0.1242,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.718680586406042e-05,
+      "loss": 0.207,
+      "step": 2545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.718125277654376e-05,
+      "loss": 0.396,
+      "step": 2550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7175699689027102e-05,
+      "loss": 0.0785,
+      "step": 2555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.717014660151044e-05,
+      "loss": 0.1103,
+      "step": 2560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.716459351399378e-05,
+      "loss": 0.084,
+      "step": 2565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7159040426477124e-05,
+      "loss": 0.1304,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7153487338960463e-05,
+      "loss": 0.0576,
+      "step": 2575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7147934251443806e-05,
+      "loss": 0.0594,
+      "step": 2580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7142381163927145e-05,
+      "loss": 0.0975,
+      "step": 2585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7136828076410485e-05,
+      "loss": 0.1714,
+      "step": 2590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7131274988893827e-05,
+      "loss": 0.1486,
+      "step": 2595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7125721901377167e-05,
+      "loss": 0.3571,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7120168813860506e-05,
+      "loss": 0.1006,
+      "step": 2605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.711461572634385e-05,
+      "loss": 0.1108,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7109062638827188e-05,
+      "loss": 0.1039,
+      "step": 2615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.710350955131053e-05,
+      "loss": 0.117,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.709795646379387e-05,
+      "loss": 0.1117,
+      "step": 2625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.709240337627721e-05,
+      "loss": 0.0795,
+      "step": 2630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7086850288760552e-05,
+      "loss": 0.0756,
+      "step": 2635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7081297201243895e-05,
+      "loss": 0.1152,
+      "step": 2640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.707574411372723e-05,
+      "loss": 0.2112,
+      "step": 2645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7070191026210574e-05,
+      "loss": 0.3646,
+      "step": 2650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7064637938693917e-05,
+      "loss": 0.078,
+      "step": 2655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7059084851177256e-05,
+      "loss": 0.0902,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7053531763660595e-05,
+      "loss": 0.0971,
+      "step": 2665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7047978676143938e-05,
+      "loss": 0.1079,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7042425588627278e-05,
+      "loss": 0.061,
+      "step": 2675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.703687250111062e-05,
+      "loss": 0.0765,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.703131941359396e-05,
+      "loss": 0.1012,
+      "step": 2685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.70257663260773e-05,
+      "loss": 0.0899,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7020213238560642e-05,
+      "loss": 0.1788,
+      "step": 2695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.701466015104398e-05,
+      "loss": 0.2892,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7009107063527324e-05,
+      "loss": 0.0653,
+      "step": 2705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7003553976010663e-05,
+      "loss": 0.0878,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6998000888494003e-05,
+      "loss": 0.1398,
+      "step": 2715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6992447800977345e-05,
+      "loss": 0.1062,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6986894713460685e-05,
+      "loss": 0.0806,
+      "step": 2725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6981341625944024e-05,
+      "loss": 0.0911,
+      "step": 2730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6975788538427367e-05,
+      "loss": 0.082,
+      "step": 2735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.697023545091071e-05,
+      "loss": 0.1154,
+      "step": 2740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.696468236339405e-05,
+      "loss": 0.1959,
+      "step": 2745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.695912927587739e-05,
+      "loss": 0.367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.695357618836073e-05,
+      "loss": 0.0805,
+      "step": 2755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.694802310084407e-05,
+      "loss": 0.1069,
+      "step": 2760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6942470013327413e-05,
+      "loss": 0.1016,
+      "step": 2765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6936916925810753e-05,
+      "loss": 0.1137,
+      "step": 2770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6931363838294092e-05,
+      "loss": 0.0935,
+      "step": 2775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6925810750777435e-05,
+      "loss": 0.0761,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6920257663260774e-05,
+      "loss": 0.0943,
+      "step": 2785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6914704575744114e-05,
+      "loss": 0.1072,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6909151488227456e-05,
+      "loss": 0.2146,
+      "step": 2795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6903598400710796e-05,
+      "loss": 0.3725,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.689804531319414e-05,
+      "loss": 0.092,
+      "step": 2805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6892492225677478e-05,
+      "loss": 0.0832,
+      "step": 2810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6886939138160817e-05,
+      "loss": 0.1009,
+      "step": 2815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.688138605064416e-05,
+      "loss": 0.0979,
+      "step": 2820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.68758329631275e-05,
+      "loss": 0.0872,
+      "step": 2825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.687027987561084e-05,
+      "loss": 0.1036,
+      "step": 2830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.686472678809418e-05,
+      "loss": 0.0925,
+      "step": 2835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.685917370057752e-05,
+      "loss": 0.145,
+      "step": 2840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6853620613060864e-05,
+      "loss": 0.1965,
+      "step": 2845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6848067525544206e-05,
+      "loss": 0.3851,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6842514438027546e-05,
+      "loss": 0.1303,
+      "step": 2855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6836961350510885e-05,
+      "loss": 0.0944,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6831408262994228e-05,
+      "loss": 0.1084,
+      "step": 2865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6825855175477567e-05,
+      "loss": 0.1098,
+      "step": 2870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6820302087960907e-05,
+      "loss": 0.1134,
+      "step": 2875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.681474900044425e-05,
+      "loss": 0.0863,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.680919591292759e-05,
+      "loss": 0.0932,
+      "step": 2885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.680364282541093e-05,
+      "loss": 0.1401,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.679808973789427e-05,
+      "loss": 0.1387,
+      "step": 2895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.679253665037761e-05,
+      "loss": 0.2855,
+      "step": 2900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6786983562860953e-05,
+      "loss": 0.0765,
+      "step": 2905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6781430475344292e-05,
+      "loss": 0.1152,
+      "step": 2910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.677587738782763e-05,
+      "loss": 0.0868,
+      "step": 2915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6770324300310974e-05,
+      "loss": 0.0992,
+      "step": 2920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6764771212794314e-05,
+      "loss": 0.091,
+      "step": 2925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6759218125277657e-05,
+      "loss": 0.0412,
+      "step": 2930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6753665037760996e-05,
+      "loss": 0.0889,
+      "step": 2935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6748111950244335e-05,
+      "loss": 0.1182,
+      "step": 2940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6742558862727678e-05,
+      "loss": 0.1759,
+      "step": 2945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.673700577521102e-05,
+      "loss": 0.3003,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6731452687694357e-05,
+      "loss": 0.1007,
+      "step": 2955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.67258996001777e-05,
+      "loss": 0.0859,
+      "step": 2960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6720346512661042e-05,
+      "loss": 0.0887,
+      "step": 2965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.671479342514438e-05,
+      "loss": 0.1244,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6709240337627724e-05,
+      "loss": 0.0652,
+      "step": 2975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6703687250111064e-05,
+      "loss": 0.0617,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6698134162594403e-05,
+      "loss": 0.1101,
+      "step": 2985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6692581075077746e-05,
+      "loss": 0.1038,
+      "step": 2990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6687027987561085e-05,
+      "loss": 0.1654,
+      "step": 2995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6681474900044425e-05,
+      "loss": 0.2911,
+      "step": 3000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6675921812527767e-05,
+      "loss": 0.077,
+      "step": 3005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6670368725011107e-05,
+      "loss": 0.0814,
+      "step": 3010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.666481563749445e-05,
+      "loss": 0.1077,
+      "step": 3015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.665926254997779e-05,
+      "loss": 0.1065,
+      "step": 3020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6653709462461128e-05,
+      "loss": 0.0641,
+      "step": 3025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.664815637494447e-05,
+      "loss": 0.0768,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.664260328742781e-05,
+      "loss": 0.096,
+      "step": 3035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.663705019991115e-05,
+      "loss": 0.0943,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6631497112394492e-05,
+      "loss": 0.1964,
+      "step": 3045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6625944024877835e-05,
+      "loss": 0.2383,
+      "step": 3050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6620390937361175e-05,
+      "loss": 0.1128,
+      "step": 3055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6614837849844514e-05,
+      "loss": 0.0922,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6609284762327857e-05,
+      "loss": 0.0897,
+      "step": 3065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6603731674811196e-05,
+      "loss": 0.1518,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.659817858729454e-05,
+      "loss": 0.0945,
+      "step": 3075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6592625499777878e-05,
+      "loss": 0.0477,
+      "step": 3080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6587072412261218e-05,
+      "loss": 0.112,
+      "step": 3085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.658151932474456e-05,
+      "loss": 0.1762,
+      "step": 3090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.65759662372279e-05,
+      "loss": 0.1103,
+      "step": 3095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.657041314971124e-05,
+      "loss": 0.3118,
+      "step": 3100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6564860062194582e-05,
+      "loss": 0.1042,
+      "step": 3105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.655930697467792e-05,
+      "loss": 0.0835,
+      "step": 3110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6553753887161264e-05,
+      "loss": 0.0869,
+      "step": 3115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6548200799644603e-05,
+      "loss": 0.0982,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6542647712127943e-05,
+      "loss": 0.1103,
+      "step": 3125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6537094624611285e-05,
+      "loss": 0.0477,
+      "step": 3130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6531541537094625e-05,
+      "loss": 0.0709,
+      "step": 3135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6525988449577968e-05,
+      "loss": 0.0971,
+      "step": 3140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6520435362061307e-05,
+      "loss": 0.1319,
+      "step": 3145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6514882274544646e-05,
+      "loss": 0.2376,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.650932918702799e-05,
+      "loss": 0.0834,
+      "step": 3155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6503776099511332e-05,
+      "loss": 0.0953,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.649822301199467e-05,
+      "loss": 0.1127,
+      "step": 3165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.649266992447801e-05,
+      "loss": 0.1344,
+      "step": 3170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6487116836961353e-05,
+      "loss": 0.0798,
+      "step": 3175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6481563749444693e-05,
+      "loss": 0.0569,
+      "step": 3180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6476010661928032e-05,
+      "loss": 0.0954,
+      "step": 3185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6470457574411375e-05,
+      "loss": 0.0844,
+      "step": 3190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6464904486894714e-05,
+      "loss": 0.1709,
+      "step": 3195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6459351399378057e-05,
+      "loss": 0.3021,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6453798311861396e-05,
+      "loss": 0.0801,
+      "step": 3205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6448245224344736e-05,
+      "loss": 0.099,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.644269213682808e-05,
+      "loss": 0.1008,
+      "step": 3215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6437139049311418e-05,
+      "loss": 0.1072,
+      "step": 3220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6431585961794757e-05,
+      "loss": 0.0853,
+      "step": 3225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.64260328742781e-05,
+      "loss": 0.0733,
+      "step": 3230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.642047978676144e-05,
+      "loss": 0.0471,
+      "step": 3235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6414926699244782e-05,
+      "loss": 0.1344,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6409373611728125e-05,
+      "loss": 0.0853,
+      "step": 3245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.640382052421146e-05,
+      "loss": 0.2943,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6398267436694804e-05,
+      "loss": 0.0834,
+      "step": 3255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6392714349178146e-05,
+      "loss": 0.0791,
+      "step": 3260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6387161261661486e-05,
+      "loss": 0.1087,
+      "step": 3265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6381608174144825e-05,
+      "loss": 0.1174,
+      "step": 3270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6376055086628168e-05,
+      "loss": 0.1023,
+      "step": 3275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6370501999111507e-05,
+      "loss": 0.0464,
+      "step": 3280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.636494891159485e-05,
+      "loss": 0.0904,
+      "step": 3285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.635939582407819e-05,
+      "loss": 0.1299,
+      "step": 3290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.635384273656153e-05,
+      "loss": 0.2206,
+      "step": 3295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.634828964904487e-05,
+      "loss": 0.4083,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.634273656152821e-05,
+      "loss": 0.0842,
+      "step": 3305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.633718347401155e-05,
+      "loss": 0.1296,
+      "step": 3310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6331630386494893e-05,
+      "loss": 0.0928,
+      "step": 3315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6326077298978232e-05,
+      "loss": 0.1003,
+      "step": 3320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6320524211461575e-05,
+      "loss": 0.1134,
+      "step": 3325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6314971123944914e-05,
+      "loss": 0.0904,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6309418036428254e-05,
+      "loss": 0.058,
+      "step": 3335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6303864948911597e-05,
+      "loss": 0.1375,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6298311861394936e-05,
+      "loss": 0.1815,
+      "step": 3345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6292758773878275e-05,
+      "loss": 0.3636,
+      "step": 3350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6287205686361618e-05,
+      "loss": 0.071,
+      "step": 3355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.628165259884496e-05,
+      "loss": 0.1088,
+      "step": 3360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.62760995113283e-05,
+      "loss": 0.0964,
+      "step": 3365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.627054642381164e-05,
+      "loss": 0.0859,
+      "step": 3370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6264993336294982e-05,
+      "loss": 0.0795,
+      "step": 3375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625944024877832e-05,
+      "loss": 0.0648,
+      "step": 3380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6253887161261664e-05,
+      "loss": 0.0484,
+      "step": 3385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6248334073745004e-05,
+      "loss": 0.1102,
+      "step": 3390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6242780986228343e-05,
+      "loss": 0.1527,
+      "step": 3395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6237227898711686e-05,
+      "loss": 0.2897,
+      "step": 3400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6231674811195025e-05,
+      "loss": 0.0869,
+      "step": 3405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6226121723678368e-05,
+      "loss": 0.1021,
+      "step": 3410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6220568636161707e-05,
+      "loss": 0.124,
+      "step": 3415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6215015548645047e-05,
+      "loss": 0.158,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620946246112839e-05,
+      "loss": 0.0583,
+      "step": 3425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620390937361173e-05,
+      "loss": 0.0587,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6198356286095068e-05,
+      "loss": 0.1364,
+      "step": 3435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619280319857841e-05,
+      "loss": 0.1337,
+      "step": 3440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618725011106175e-05,
+      "loss": 0.2039,
+      "step": 3445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6181697023545093e-05,
+      "loss": 0.4012,
+      "step": 3450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6176143936028433e-05,
+      "loss": 0.0968,
+      "step": 3455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6170590848511772e-05,
+      "loss": 0.0984,
+      "step": 3460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6165037760995115e-05,
+      "loss": 0.1022,
+      "step": 3465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6159484673478457e-05,
+      "loss": 0.1275,
+      "step": 3470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6153931585961797e-05,
+      "loss": 0.1107,
+      "step": 3475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6148378498445136e-05,
+      "loss": 0.0547,
+      "step": 3480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.614282541092848e-05,
+      "loss": 0.0904,
+      "step": 3485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6137272323411818e-05,
+      "loss": 0.0762,
+      "step": 3490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6131719235895158e-05,
+      "loss": 0.177,
+      "step": 3495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.61261661483785e-05,
+      "loss": 0.2028,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.612061306086184e-05,
+      "loss": 0.0763,
+      "step": 3505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6115059973345182e-05,
+      "loss": 0.0998,
+      "step": 3510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6109506885828522e-05,
+      "loss": 0.1065,
+      "step": 3515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.610395379831186e-05,
+      "loss": 0.0991,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6098400710795204e-05,
+      "loss": 0.0556,
+      "step": 3525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6092847623278543e-05,
+      "loss": 0.0796,
+      "step": 3530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6087294535761886e-05,
+      "loss": 0.0829,
+      "step": 3535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6081741448245225e-05,
+      "loss": 0.0953,
+      "step": 3540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6076188360728565e-05,
+      "loss": 0.1856,
+      "step": 3545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6070635273211908e-05,
+      "loss": 0.3222,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606508218569525e-05,
+      "loss": 0.0845,
+      "step": 3555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6059529098178586e-05,
+      "loss": 0.0808,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.605397601066193e-05,
+      "loss": 0.1092,
+      "step": 3565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6048422923145272e-05,
+      "loss": 0.0927,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.604286983562861e-05,
+      "loss": 0.112,
+      "step": 3575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.603731674811195e-05,
+      "loss": 0.0455,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6031763660595293e-05,
+      "loss": 0.0877,
+      "step": 3585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6026210573078633e-05,
+      "loss": 0.1117,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6020657485561975e-05,
+      "loss": 0.1989,
+      "step": 3595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6015104398045315e-05,
+      "loss": 0.3288,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6009551310528654e-05,
+      "loss": 0.0716,
+      "step": 3605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6003998223011997e-05,
+      "loss": 0.0858,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5998445135495336e-05,
+      "loss": 0.0821,
+      "step": 3615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5992892047978676e-05,
+      "loss": 0.1144,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.598733896046202e-05,
+      "loss": 0.0889,
+      "step": 3625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5981785872945358e-05,
+      "loss": 0.0663,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.59762327854287e-05,
+      "loss": 0.1213,
+      "step": 3635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.597067969791204e-05,
+      "loss": 0.1056,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.596512661039538e-05,
+      "loss": 0.1649,
+      "step": 3645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5959573522878722e-05,
+      "loss": 0.3544,
+      "step": 3650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.595402043536206e-05,
+      "loss": 0.0892,
+      "step": 3655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5948467347845404e-05,
+      "loss": 0.0998,
+      "step": 3660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5942914260328744e-05,
+      "loss": 0.0716,
+      "step": 3665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5937361172812086e-05,
+      "loss": 0.0938,
+      "step": 3670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5931808085295426e-05,
+      "loss": 0.0911,
+      "step": 3675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.592625499777877e-05,
+      "loss": 0.0515,
+      "step": 3680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5920701910262108e-05,
+      "loss": 0.0547,
+      "step": 3685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5915148822745447e-05,
+      "loss": 0.1367,
+      "step": 3690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.590959573522879e-05,
+      "loss": 0.184,
+      "step": 3695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.590404264771213e-05,
+      "loss": 0.4413,
+      "step": 3700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.589848956019547e-05,
+      "loss": 0.079,
+      "step": 3705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.589293647267881e-05,
+      "loss": 0.0785,
+      "step": 3710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.588738338516215e-05,
+      "loss": 0.079,
+      "step": 3715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5881830297645494e-05,
+      "loss": 0.1071,
+      "step": 3720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5876277210128833e-05,
+      "loss": 0.0562,
+      "step": 3725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5870724122612172e-05,
+      "loss": 0.0541,
+      "step": 3730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5865171035095515e-05,
+      "loss": 0.0508,
+      "step": 3735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5859617947578854e-05,
+      "loss": 0.087,
+      "step": 3740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5854064860062194e-05,
+      "loss": 0.1332,
+      "step": 3745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5848511772545537e-05,
+      "loss": 0.3807,
+      "step": 3750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5842958685028876e-05,
+      "loss": 0.0847,
+      "step": 3755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.583740559751222e-05,
+      "loss": 0.1234,
+      "step": 3760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5831852509995558e-05,
+      "loss": 0.1025,
+      "step": 3765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5826299422478897e-05,
+      "loss": 0.104,
+      "step": 3770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.582074633496224e-05,
+      "loss": 0.1179,
+      "step": 3775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5815193247445583e-05,
+      "loss": 0.0519,
+      "step": 3780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5809640159928922e-05,
+      "loss": 0.0564,
+      "step": 3785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.580408707241226e-05,
+      "loss": 0.0819,
+      "step": 3790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5798533984895604e-05,
+      "loss": 0.2076,
+      "step": 3795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5792980897378944e-05,
+      "loss": 0.4112,
+      "step": 3800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5787427809862287e-05,
+      "loss": 0.101,
+      "step": 3805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5781874722345626e-05,
+      "loss": 0.0948,
+      "step": 3810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5776321634828965e-05,
+      "loss": 0.0874,
+      "step": 3815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5770768547312308e-05,
+      "loss": 0.0904,
+      "step": 3820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5765215459795647e-05,
+      "loss": 0.0897,
+      "step": 3825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5759662372278987e-05,
+      "loss": 0.0367,
+      "step": 3830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.575410928476233e-05,
+      "loss": 0.0713,
+      "step": 3835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.574855619724567e-05,
+      "loss": 0.1449,
+      "step": 3840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.574300310972901e-05,
+      "loss": 0.1427,
+      "step": 3845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.573745002221235e-05,
+      "loss": 0.2691,
+      "step": 3850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.573189693469569e-05,
+      "loss": 0.1211,
+      "step": 3855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5726343847179033e-05,
+      "loss": 0.0944,
+      "step": 3860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5720790759662376e-05,
+      "loss": 0.1037,
+      "step": 3865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5715237672145712e-05,
+      "loss": 0.08,
+      "step": 3870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5709684584629055e-05,
+      "loss": 0.1048,
+      "step": 3875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5704131497112397e-05,
+      "loss": 0.0437,
+      "step": 3880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5698578409595737e-05,
+      "loss": 0.0731,
+      "step": 3885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5693025322079076e-05,
+      "loss": 0.1378,
+      "step": 3890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.568747223456242e-05,
+      "loss": 0.1637,
+      "step": 3895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5681919147045758e-05,
+      "loss": 0.2737,
+      "step": 3900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.56763660595291e-05,
+      "loss": 0.0846,
+      "step": 3905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.567081297201244e-05,
+      "loss": 0.0757,
+      "step": 3910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.566525988449578e-05,
+      "loss": 0.09,
+      "step": 3915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5659706796979122e-05,
+      "loss": 0.1917,
+      "step": 3920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5654153709462462e-05,
+      "loss": 0.0921,
+      "step": 3925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5648600621945805e-05,
+      "loss": 0.0803,
+      "step": 3930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5643047534429144e-05,
+      "loss": 0.1142,
+      "step": 3935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5637494446912483e-05,
+      "loss": 0.1233,
+      "step": 3940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5631941359395826e-05,
+      "loss": 0.1779,
+      "step": 3945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5626388271879165e-05,
+      "loss": 0.305,
+      "step": 3950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5620835184362505e-05,
+      "loss": 0.1212,
+      "step": 3955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5615282096845848e-05,
+      "loss": 0.0716,
+      "step": 3960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5609729009329187e-05,
+      "loss": 0.0825,
+      "step": 3965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.560417592181253e-05,
+      "loss": 0.13,
+      "step": 3970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.559862283429587e-05,
+      "loss": 0.0773,
+      "step": 3975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5593069746779212e-05,
+      "loss": 0.0568,
+      "step": 3980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.558751665926255e-05,
+      "loss": 0.0973,
+      "step": 3985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5581963571745894e-05,
+      "loss": 0.1212,
+      "step": 3990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5576410484229233e-05,
+      "loss": 0.2039,
+      "step": 3995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5570857396712573e-05,
+      "loss": 0.2844,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.039118871092796326,
+      "eval_runtime": 1816.348,
+      "eval_samples_per_second": 8.819,
+      "eval_steps_per_second": 2.205,
+      "eval_wer": 0.10885322755138839,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5565304309195915e-05,
+      "loss": 0.0808,
+      "step": 4005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5559751221679255e-05,
+      "loss": 0.0893,
+      "step": 4010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5554198134162594e-05,
+      "loss": 0.1097,
+      "step": 4015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5548645046645937e-05,
+      "loss": 0.1095,
+      "step": 4020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5543091959129276e-05,
+      "loss": 0.0481,
+      "step": 4025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.553753887161262e-05,
+      "loss": 0.053,
+      "step": 4030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.553198578409596e-05,
+      "loss": 0.0695,
+      "step": 4035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5526432696579298e-05,
+      "loss": 0.0851,
+      "step": 4040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.552087960906264e-05,
+      "loss": 0.2526,
+      "step": 4045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.551532652154598e-05,
+      "loss": 0.2974,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5509773434029323e-05,
+      "loss": 0.0649,
+      "step": 4055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5504220346512662e-05,
+      "loss": 0.083,
+      "step": 4060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5498667258996e-05,
+      "loss": 0.1154,
+      "step": 4065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5493114171479344e-05,
+      "loss": 0.1205,
+      "step": 4070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5487561083962687e-05,
+      "loss": 0.0725,
+      "step": 4075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5482007996446023e-05,
+      "loss": 0.0479,
+      "step": 4080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5476454908929366e-05,
+      "loss": 0.0567,
+      "step": 4085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.547090182141271e-05,
+      "loss": 0.0601,
+      "step": 4090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5465348733896048e-05,
+      "loss": 0.1391,
+      "step": 4095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5459795646379387e-05,
+      "loss": 0.2792,
+      "step": 4100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.545424255886273e-05,
+      "loss": 0.0962,
+      "step": 4105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.544868947134607e-05,
+      "loss": 0.095,
+      "step": 4110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5443136383829412e-05,
+      "loss": 0.093,
+      "step": 4115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.543758329631275e-05,
+      "loss": 0.0956,
+      "step": 4120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.543203020879609e-05,
+      "loss": 0.0835,
+      "step": 4125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5426477121279434e-05,
+      "loss": 0.0798,
+      "step": 4130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5420924033762773e-05,
+      "loss": 0.1073,
+      "step": 4135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5415370946246112e-05,
+      "loss": 0.106,
+      "step": 4140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5409817858729455e-05,
+      "loss": 0.2011,
+      "step": 4145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5404264771212794e-05,
+      "loss": 0.4367,
+      "step": 4150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5398711683696137e-05,
+      "loss": 0.0719,
+      "step": 4155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5393158596179477e-05,
+      "loss": 0.0725,
+      "step": 4160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5387605508662816e-05,
+      "loss": 0.1205,
+      "step": 4165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.538205242114616e-05,
+      "loss": 0.0822,
+      "step": 4170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.53764993336295e-05,
+      "loss": 0.0884,
+      "step": 4175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5370946246112837e-05,
+      "loss": 0.0544,
+      "step": 4180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.536539315859618e-05,
+      "loss": 0.0357,
+      "step": 4185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5359840071079523e-05,
+      "loss": 0.1223,
+      "step": 4190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5354286983562862e-05,
+      "loss": 0.156,
+      "step": 4195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5348733896046205e-05,
+      "loss": 0.3048,
+      "step": 4200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5343180808529544e-05,
+      "loss": 0.0934,
+      "step": 4205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5337627721012884e-05,
+      "loss": 0.1207,
+      "step": 4210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5332074633496227e-05,
+      "loss": 0.0921,
+      "step": 4215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5326521545979566e-05,
+      "loss": 0.1101,
+      "step": 4220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5320968458462905e-05,
+      "loss": 0.0986,
+      "step": 4225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5315415370946248e-05,
+      "loss": 0.0326,
+      "step": 4230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5309862283429587e-05,
+      "loss": 0.0517,
+      "step": 4235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.530430919591293e-05,
+      "loss": 0.0859,
+      "step": 4240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.529875610839627e-05,
+      "loss": 0.1802,
+      "step": 4245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.529320302087961e-05,
+      "loss": 0.3513,
+      "step": 4250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.528764993336295e-05,
+      "loss": 0.0682,
+      "step": 4255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.528209684584629e-05,
+      "loss": 0.1053,
+      "step": 4260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.527654375832963e-05,
+      "loss": 0.1228,
+      "step": 4265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5270990670812973e-05,
+      "loss": 0.1026,
+      "step": 4270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5265437583296313e-05,
+      "loss": 0.0545,
+      "step": 4275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5259884495779655e-05,
+      "loss": 0.0536,
+      "step": 4280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5254331408262995e-05,
+      "loss": 0.0977,
+      "step": 4285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5248778320746336e-05,
+      "loss": 0.1605,
+      "step": 4290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5243225233229677e-05,
+      "loss": 0.1559,
+      "step": 4295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5237672145713018e-05,
+      "loss": 0.3257,
+      "step": 4300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5232119058196357e-05,
+      "loss": 0.0931,
+      "step": 4305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5226565970679698e-05,
+      "loss": 0.0717,
+      "step": 4310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.522101288316304e-05,
+      "loss": 0.0744,
+      "step": 4315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.521545979564638e-05,
+      "loss": 0.1406,
+      "step": 4320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5209906708129723e-05,
+      "loss": 0.0879,
+      "step": 4325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.520435362061306e-05,
+      "loss": 0.0413,
+      "step": 4330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5198800533096404e-05,
+      "loss": 0.0539,
+      "step": 4335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5193247445579745e-05,
+      "loss": 0.087,
+      "step": 4340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5187694358063086e-05,
+      "loss": 0.1241,
+      "step": 4345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5182141270546425e-05,
+      "loss": 0.2865,
+      "step": 4350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5176588183029766e-05,
+      "loss": 0.0835,
+      "step": 4355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5171035095513107e-05,
+      "loss": 0.0901,
+      "step": 4360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5165482007996448e-05,
+      "loss": 0.1145,
+      "step": 4365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5159928920479788e-05,
+      "loss": 0.1068,
+      "step": 4370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5154375832963129e-05,
+      "loss": 0.0441,
+      "step": 4375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.514882274544647e-05,
+      "loss": 0.0632,
+      "step": 4380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.514326965792981e-05,
+      "loss": 0.1224,
+      "step": 4385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.513771657041315e-05,
+      "loss": 0.1325,
+      "step": 4390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5132163482896491e-05,
+      "loss": 0.1927,
+      "step": 4395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5126610395379832e-05,
+      "loss": 0.2885,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5121057307863173e-05,
+      "loss": 0.0798,
+      "step": 4405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5115504220346513e-05,
+      "loss": 0.087,
+      "step": 4410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5109951132829854e-05,
+      "loss": 0.0931,
+      "step": 4415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5104398045313195e-05,
+      "loss": 0.0984,
+      "step": 4420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5098844957796538e-05,
+      "loss": 0.0624,
+      "step": 4425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5093291870279875e-05,
+      "loss": 0.0547,
+      "step": 4430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5087738782763216e-05,
+      "loss": 0.094,
+      "step": 4435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5082185695246559e-05,
+      "loss": 0.0959,
+      "step": 4440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.50766326077299e-05,
+      "loss": 0.1507,
+      "step": 4445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.507107952021324e-05,
+      "loss": 0.2696,
+      "step": 4450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.506552643269658e-05,
+      "loss": 0.0944,
+      "step": 4455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5059973345179922e-05,
+      "loss": 0.0857,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5054420257663263e-05,
+      "loss": 0.0834,
+      "step": 4465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5048867170146604e-05,
+      "loss": 0.1062,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5043314082629943e-05,
+      "loss": 0.0683,
+      "step": 4475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5037760995113284e-05,
+      "loss": 0.0512,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5032207907596625e-05,
+      "loss": 0.0896,
+      "step": 4485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5026654820079966e-05,
+      "loss": 0.1176,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5022212350066638e-05,
+      "loss": 0.3841,
+      "step": 4495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5016659262549979e-05,
+      "loss": 0.3718,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5011106175033318e-05,
+      "loss": 0.0946,
+      "step": 4505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.500555308751666e-05,
+      "loss": 0.058,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1082,
+      "step": 4515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4994446912483343e-05,
+      "loss": 0.1132,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4988893824966684e-05,
+      "loss": 0.0723,
+      "step": 4525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4983340737450024e-05,
+      "loss": 0.0698,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4977787649933365e-05,
+      "loss": 0.0791,
+      "step": 4535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4972234562416706e-05,
+      "loss": 0.0893,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4966681474900047e-05,
+      "loss": 0.1476,
+      "step": 4545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4961128387383386e-05,
+      "loss": 0.3097,
+      "step": 4550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4955575299866727e-05,
+      "loss": 0.0913,
+      "step": 4555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4950022212350068e-05,
+      "loss": 0.09,
+      "step": 4560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.494446912483341e-05,
+      "loss": 0.1254,
+      "step": 4565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4938916037316749e-05,
+      "loss": 0.0895,
+      "step": 4570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.493336294980009e-05,
+      "loss": 0.0573,
+      "step": 4575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.492780986228343e-05,
+      "loss": 0.049,
+      "step": 4580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4922256774766772e-05,
+      "loss": 0.0823,
+      "step": 4585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4916703687250111e-05,
+      "loss": 0.123,
+      "step": 4590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4911150599733452e-05,
+      "loss": 0.1552,
+      "step": 4595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4905597512216793e-05,
+      "loss": 0.4451,
+      "step": 4600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4900044424700134e-05,
+      "loss": 0.1121,
+      "step": 4605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4894491337183474e-05,
+      "loss": 0.0816,
+      "step": 4610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4888938249666815e-05,
+      "loss": 0.0919,
+      "step": 4615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4883385162150158e-05,
+      "loss": 0.1133,
+      "step": 4620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4877832074633499e-05,
+      "loss": 0.0803,
+      "step": 4625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4872278987116836e-05,
+      "loss": 0.0438,
+      "step": 4630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4866725899600179e-05,
+      "loss": 0.0981,
+      "step": 4635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.486117281208352e-05,
+      "loss": 0.0885,
+      "step": 4640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4855619724566861e-05,
+      "loss": 0.2147,
+      "step": 4645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.48500666370502e-05,
+      "loss": 0.2633,
+      "step": 4650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4844513549533542e-05,
+      "loss": 0.0868,
+      "step": 4655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4838960462016883e-05,
+      "loss": 0.0864,
+      "step": 4660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4833407374500224e-05,
+      "loss": 0.0768,
+      "step": 4665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4827854286983565e-05,
+      "loss": 0.0965,
+      "step": 4670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4822301199466904e-05,
+      "loss": 0.0779,
+      "step": 4675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4816748111950245e-05,
+      "loss": 0.0463,
+      "step": 4680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4811195024433586e-05,
+      "loss": 0.0687,
+      "step": 4685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4805641936916927e-05,
+      "loss": 0.1446,
+      "step": 4690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4800088849400267e-05,
+      "loss": 0.2321,
+      "step": 4695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4794535761883608e-05,
+      "loss": 0.2379,
+      "step": 4700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4788982674366949e-05,
+      "loss": 0.0683,
+      "step": 4705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4783429586850292e-05,
+      "loss": 0.0992,
+      "step": 4710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.477787649933363e-05,
+      "loss": 0.1148,
+      "step": 4715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.477232341181697e-05,
+      "loss": 0.0855,
+      "step": 4720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4766770324300313e-05,
+      "loss": 0.0998,
+      "step": 4725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4761217236783654e-05,
+      "loss": 0.042,
+      "step": 4730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4755664149266994e-05,
+      "loss": 0.0834,
+      "step": 4735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4750111061750335e-05,
+      "loss": 0.0889,
+      "step": 4740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4744557974233676e-05,
+      "loss": 0.1453,
+      "step": 4745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4739004886717017e-05,
+      "loss": 0.3652,
+      "step": 4750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4733451799200356e-05,
+      "loss": 0.091,
+      "step": 4755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4727898711683697e-05,
+      "loss": 0.0837,
+      "step": 4760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4722345624167038e-05,
+      "loss": 0.0931,
+      "step": 4765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.471679253665038e-05,
+      "loss": 0.1036,
+      "step": 4770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4711239449133719e-05,
+      "loss": 0.0945,
+      "step": 4775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.470568636161706e-05,
+      "loss": 0.0556,
+      "step": 4780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.47001332741004e-05,
+      "loss": 0.1127,
+      "step": 4785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4694580186583742e-05,
+      "loss": 0.11,
+      "step": 4790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4689027099067083e-05,
+      "loss": 0.1356,
+      "step": 4795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4683474011550422e-05,
+      "loss": 0.3402,
+      "step": 4800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4677920924033763e-05,
+      "loss": 0.0831,
+      "step": 4805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4672367836517104e-05,
+      "loss": 0.0801,
+      "step": 4810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4666814749000447e-05,
+      "loss": 0.0765,
+      "step": 4815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4661261661483785e-05,
+      "loss": 0.0971,
+      "step": 4820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4655708573967128e-05,
+      "loss": 0.0931,
+      "step": 4825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4650155486450469e-05,
+      "loss": 0.0386,
+      "step": 4830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.464460239893381e-05,
+      "loss": 0.0818,
+      "step": 4835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4639049311417149e-05,
+      "loss": 0.122,
+      "step": 4840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.463349622390049e-05,
+      "loss": 0.1686,
+      "step": 4845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4627943136383831e-05,
+      "loss": 0.3814,
+      "step": 4850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4622390048867172e-05,
+      "loss": 0.1258,
+      "step": 4855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4616836961350512e-05,
+      "loss": 0.0801,
+      "step": 4860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4611283873833853e-05,
+      "loss": 0.1104,
+      "step": 4865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4605730786317194e-05,
+      "loss": 0.1128,
+      "step": 4870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4600177698800535e-05,
+      "loss": 0.0878,
+      "step": 4875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4594624611283874e-05,
+      "loss": 0.0973,
+      "step": 4880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4589071523767215e-05,
+      "loss": 0.1013,
+      "step": 4885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4583518436250556e-05,
+      "loss": 0.0862,
+      "step": 4890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4577965348733897e-05,
+      "loss": 0.1651,
+      "step": 4895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4572412261217237e-05,
+      "loss": 0.3062,
+      "step": 4900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4566859173700578e-05,
+      "loss": 0.0759,
+      "step": 4905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4561306086183919e-05,
+      "loss": 0.085,
+      "step": 4910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.455575299866726e-05,
+      "loss": 0.0755,
+      "step": 4915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.45501999111506e-05,
+      "loss": 0.1068,
+      "step": 4920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.454464682363394e-05,
+      "loss": 0.075,
+      "step": 4925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4539093736117283e-05,
+      "loss": 0.064,
+      "step": 4930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4533540648600624e-05,
+      "loss": 0.138,
+      "step": 4935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4527987561083965e-05,
+      "loss": 0.0975,
+      "step": 4940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4522434473567305e-05,
+      "loss": 0.1526,
+      "step": 4945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4516881386050646e-05,
+      "loss": 0.3765,
+      "step": 4950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4511328298533987e-05,
+      "loss": 0.0871,
+      "step": 4955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4505775211017328e-05,
+      "loss": 0.0829,
+      "step": 4960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4500222123500667e-05,
+      "loss": 0.0882,
+      "step": 4965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4494669035984008e-05,
+      "loss": 0.1186,
+      "step": 4970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.448911594846735e-05,
+      "loss": 0.0602,
+      "step": 4975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.448356286095069e-05,
+      "loss": 0.0834,
+      "step": 4980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.447800977343403e-05,
+      "loss": 0.0827,
+      "step": 4985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.447245668591737e-05,
+      "loss": 0.0868,
+      "step": 4990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4466903598400712e-05,
+      "loss": 0.205,
+      "step": 4995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4461350510884053e-05,
+      "loss": 0.1816,
+      "step": 5000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4455797423367392e-05,
+      "loss": 0.091,
+      "step": 5005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4450244335850733e-05,
+      "loss": 0.1106,
+      "step": 5010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4444691248334074e-05,
+      "loss": 0.0741,
+      "step": 5015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4439138160817417e-05,
+      "loss": 0.0894,
+      "step": 5020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4433585073300755e-05,
+      "loss": 0.078,
+      "step": 5025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4428031985784096e-05,
+      "loss": 0.0711,
+      "step": 5030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4422478898267439e-05,
+      "loss": 0.0569,
+      "step": 5035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.441692581075078e-05,
+      "loss": 0.132,
+      "step": 5040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4411372723234119e-05,
+      "loss": 0.2066,
+      "step": 5045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.440581963571746e-05,
+      "loss": 0.3894,
+      "step": 5050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4400266548200801e-05,
+      "loss": 0.074,
+      "step": 5055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4394713460684142e-05,
+      "loss": 0.0821,
+      "step": 5060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4389160373167483e-05,
+      "loss": 0.0926,
+      "step": 5065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4383607285650823e-05,
+      "loss": 0.1133,
+      "step": 5070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4378054198134164e-05,
+      "loss": 0.0966,
+      "step": 5075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4372501110617505e-05,
+      "loss": 0.0571,
+      "step": 5080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4366948023100846e-05,
+      "loss": 0.0918,
+      "step": 5085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4361394935584185e-05,
+      "loss": 0.1036,
+      "step": 5090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4355841848067526e-05,
+      "loss": 0.1529,
+      "step": 5095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4350288760550867e-05,
+      "loss": 0.3466,
+      "step": 5100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4344735673034208e-05,
+      "loss": 0.0792,
+      "step": 5105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4339182585517548e-05,
+      "loss": 0.0851,
+      "step": 5110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4333629498000889e-05,
+      "loss": 0.0791,
+      "step": 5115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.432807641048423e-05,
+      "loss": 0.1158,
+      "step": 5120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4322523322967573e-05,
+      "loss": 0.0901,
+      "step": 5125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.431697023545091e-05,
+      "loss": 0.0801,
+      "step": 5130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4311417147934253e-05,
+      "loss": 0.1065,
+      "step": 5135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4305864060417594e-05,
+      "loss": 0.0967,
+      "step": 5140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4300310972900935e-05,
+      "loss": 0.1442,
+      "step": 5145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4294757885384275e-05,
+      "loss": 0.3527,
+      "step": 5150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4289204797867616e-05,
+      "loss": 0.1082,
+      "step": 5155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4283651710350957e-05,
+      "loss": 0.0742,
+      "step": 5160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4278098622834298e-05,
+      "loss": 0.081,
+      "step": 5165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4272545535317637e-05,
+      "loss": 0.1064,
+      "step": 5170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4266992447800978e-05,
+      "loss": 0.1017,
+      "step": 5175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.426143936028432e-05,
+      "loss": 0.0786,
+      "step": 5180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.425588627276766e-05,
+      "loss": 0.0571,
+      "step": 5185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4250333185251e-05,
+      "loss": 0.0997,
+      "step": 5190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.424478009773434e-05,
+      "loss": 0.1396,
+      "step": 5195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4239227010217682e-05,
+      "loss": 0.343,
+      "step": 5200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4233673922701023e-05,
+      "loss": 0.0666,
+      "step": 5205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4228120835184364e-05,
+      "loss": 0.0859,
+      "step": 5210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4222567747667703e-05,
+      "loss": 0.0949,
+      "step": 5215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4217014660151044e-05,
+      "loss": 0.1246,
+      "step": 5220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4211461572634385e-05,
+      "loss": 0.062,
+      "step": 5225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4205908485117728e-05,
+      "loss": 0.0696,
+      "step": 5230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4200355397601066e-05,
+      "loss": 0.1217,
+      "step": 5235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4194802310084409e-05,
+      "loss": 0.1105,
+      "step": 5240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.418924922256775e-05,
+      "loss": 0.1159,
+      "step": 5245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.418369613505109e-05,
+      "loss": 0.2965,
+      "step": 5250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.417814304753443e-05,
+      "loss": 0.0827,
+      "step": 5255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4172589960017771e-05,
+      "loss": 0.0646,
+      "step": 5260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4167036872501112e-05,
+      "loss": 0.0942,
+      "step": 5265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4161483784984453e-05,
+      "loss": 0.0923,
+      "step": 5270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4155930697467793e-05,
+      "loss": 0.0579,
+      "step": 5275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4150377609951134e-05,
+      "loss": 0.0786,
+      "step": 5280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4144824522434475e-05,
+      "loss": 0.086,
+      "step": 5285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4139271434917816e-05,
+      "loss": 0.1091,
+      "step": 5290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4133718347401155e-05,
+      "loss": 0.2407,
+      "step": 5295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4128165259884496e-05,
+      "loss": 0.3401,
+      "step": 5300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4122612172367837e-05,
+      "loss": 0.1018,
+      "step": 5305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4117059084851178e-05,
+      "loss": 0.0991,
+      "step": 5310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4111505997334518e-05,
+      "loss": 0.1278,
+      "step": 5315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4105952909817859e-05,
+      "loss": 0.1081,
+      "step": 5320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.41003998223012e-05,
+      "loss": 0.1059,
+      "step": 5325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4094846734784543e-05,
+      "loss": 0.0501,
+      "step": 5330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4089293647267884e-05,
+      "loss": 0.1164,
+      "step": 5335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4083740559751221e-05,
+      "loss": 0.1149,
+      "step": 5340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4078187472234564e-05,
+      "loss": 0.1648,
+      "step": 5345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4072634384717905e-05,
+      "loss": 0.3748,
+      "step": 5350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4067081297201246e-05,
+      "loss": 0.079,
+      "step": 5355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4061528209684586e-05,
+      "loss": 0.0679,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4055975122167927e-05,
+      "loss": 0.0832,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4050422034651268e-05,
+      "loss": 0.114,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4044868947134609e-05,
+      "loss": 0.0807,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4039315859617948e-05,
+      "loss": 0.0415,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.403376277210129e-05,
+      "loss": 0.0369,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.402820968458463e-05,
+      "loss": 0.1212,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4022656597067971e-05,
+      "loss": 0.2648,
+      "step": 5395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.401710350955131e-05,
+      "loss": 0.326,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4011550422034652e-05,
+      "loss": 0.0696,
+      "step": 5405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4005997334517993e-05,
+      "loss": 0.091,
+      "step": 5410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4000444247001334e-05,
+      "loss": 0.1021,
+      "step": 5415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3994891159484673e-05,
+      "loss": 0.1057,
+      "step": 5420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3989338071968014e-05,
+      "loss": 0.0742,
+      "step": 5425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3983784984451355e-05,
+      "loss": 0.0442,
+      "step": 5430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3978231896934698e-05,
+      "loss": 0.0694,
+      "step": 5435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3972678809418036e-05,
+      "loss": 0.0808,
+      "step": 5440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.3967125721901379e-05,
+      "loss": 0.1707,
+      "step": 5445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.396157263438472e-05,
+      "loss": 0.2175,
+      "step": 5450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.395601954686806e-05,
+      "loss": 0.0772,
+      "step": 5455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.39504664593514e-05,
+      "loss": 0.0992,
+      "step": 5460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3944913371834741e-05,
+      "loss": 0.0719,
+      "step": 5465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3939360284318082e-05,
+      "loss": 0.0915,
+      "step": 5470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3933807196801423e-05,
+      "loss": 0.0908,
+      "step": 5475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3928254109284764e-05,
+      "loss": 0.0371,
+      "step": 5480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3922701021768104e-05,
+      "loss": 0.0494,
+      "step": 5485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3917147934251445e-05,
+      "loss": 0.1084,
+      "step": 5490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3911594846734786e-05,
+      "loss": 0.14,
+      "step": 5495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3906041759218127e-05,
+      "loss": 0.3693,
+      "step": 5500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3900488671701466e-05,
+      "loss": 0.0609,
+      "step": 5505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3894935584184807e-05,
+      "loss": 0.1027,
+      "step": 5510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3889382496668148e-05,
+      "loss": 0.088,
+      "step": 5515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.388382940915149e-05,
+      "loss": 0.1229,
+      "step": 5520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3878276321634829e-05,
+      "loss": 0.0449,
+      "step": 5525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.387272323411817e-05,
+      "loss": 0.0523,
+      "step": 5530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3867170146601511e-05,
+      "loss": 0.0875,
+      "step": 5535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3861617059084854e-05,
+      "loss": 0.058,
+      "step": 5540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3856063971568191e-05,
+      "loss": 0.1768,
+      "step": 5545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3850510884051534e-05,
+      "loss": 0.3254,
+      "step": 5550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3844957796534875e-05,
+      "loss": 0.0697,
+      "step": 5555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3839404709018216e-05,
+      "loss": 0.0821,
+      "step": 5560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3833851621501556e-05,
+      "loss": 0.0736,
+      "step": 5565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3828298533984897e-05,
+      "loss": 0.0821,
+      "step": 5570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3822745446468238e-05,
+      "loss": 0.1,
+      "step": 5575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3817192358951579e-05,
+      "loss": 0.0325,
+      "step": 5580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3811639271434918e-05,
+      "loss": 0.0469,
+      "step": 5585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.380608618391826e-05,
+      "loss": 0.1042,
+      "step": 5590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.38005330964016e-05,
+      "loss": 0.1995,
+      "step": 5595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3794980008884941e-05,
+      "loss": 0.1968,
+      "step": 5600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3789426921368283e-05,
+      "loss": 0.0811,
+      "step": 5605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3783873833851622e-05,
+      "loss": 0.0966,
+      "step": 5610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3778320746334963e-05,
+      "loss": 0.1066,
+      "step": 5615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3772767658818304e-05,
+      "loss": 0.097,
+      "step": 5620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3767214571301645e-05,
+      "loss": 0.0723,
+      "step": 5625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3761661483784984e-05,
+      "loss": 0.0836,
+      "step": 5630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3756108396268325e-05,
+      "loss": 0.1059,
+      "step": 5635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3750555308751668e-05,
+      "loss": 0.096,
+      "step": 5640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.374500222123501e-05,
+      "loss": 0.1821,
+      "step": 5645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3739449133718347e-05,
+      "loss": 0.2849,
+      "step": 5650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.373389604620169e-05,
+      "loss": 0.0872,
+      "step": 5655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.372834295868503e-05,
+      "loss": 0.0803,
+      "step": 5660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3722789871168372e-05,
+      "loss": 0.088,
+      "step": 5665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3717236783651711e-05,
+      "loss": 0.0868,
+      "step": 5670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3711683696135052e-05,
+      "loss": 0.0899,
+      "step": 5675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3706130608618393e-05,
+      "loss": 0.0617,
+      "step": 5680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3700577521101734e-05,
+      "loss": 0.0903,
+      "step": 5685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3695024433585074e-05,
+      "loss": 0.1025,
+      "step": 5690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3689471346068415e-05,
+      "loss": 0.1936,
+      "step": 5695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3683918258551756e-05,
+      "loss": 0.3353,
+      "step": 5700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3678365171035097e-05,
+      "loss": 0.0707,
+      "step": 5705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3672812083518436e-05,
+      "loss": 0.0823,
+      "step": 5710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3667258996001777e-05,
+      "loss": 0.0963,
+      "step": 5715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3661705908485118e-05,
+      "loss": 0.1187,
+      "step": 5720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.365615282096846e-05,
+      "loss": 0.0632,
+      "step": 5725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3650599733451799e-05,
+      "loss": 0.0524,
+      "step": 5730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.364504664593514e-05,
+      "loss": 0.055,
+      "step": 5735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3639493558418481e-05,
+      "loss": 0.149,
+      "step": 5740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3633940470901824e-05,
+      "loss": 0.1794,
+      "step": 5745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3628387383385165e-05,
+      "loss": 0.3633,
+      "step": 5750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3622834295868504e-05,
+      "loss": 0.0882,
+      "step": 5755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3617281208351845e-05,
+      "loss": 0.0851,
+      "step": 5760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3611728120835186e-05,
+      "loss": 0.0925,
+      "step": 5765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3606175033318527e-05,
+      "loss": 0.0751,
+      "step": 5770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3600621945801867e-05,
+      "loss": 0.0654,
+      "step": 5775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3595068858285208e-05,
+      "loss": 0.045,
+      "step": 5780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3589515770768549e-05,
+      "loss": 0.0613,
+      "step": 5785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.358396268325189e-05,
+      "loss": 0.1491,
+      "step": 5790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.357840959573523e-05,
+      "loss": 0.1382,
+      "step": 5795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.357285650821857e-05,
+      "loss": 0.2285,
+      "step": 5800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3567303420701911e-05,
+      "loss": 0.1152,
+      "step": 5805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3561750333185253e-05,
+      "loss": 0.0615,
+      "step": 5810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3556197245668592e-05,
+      "loss": 0.0983,
+      "step": 5815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3550644158151933e-05,
+      "loss": 0.0896,
+      "step": 5820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3545091070635274e-05,
+      "loss": 0.08,
+      "step": 5825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3539537983118615e-05,
+      "loss": 0.048,
+      "step": 5830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3533984895601954e-05,
+      "loss": 0.0762,
+      "step": 5835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3528431808085295e-05,
+      "loss": 0.0981,
+      "step": 5840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3522878720568637e-05,
+      "loss": 0.1538,
+      "step": 5845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.351732563305198e-05,
+      "loss": 0.3178,
+      "step": 5850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3511772545535317e-05,
+      "loss": 0.0936,
+      "step": 5855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.350621945801866e-05,
+      "loss": 0.0786,
+      "step": 5860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3500666370502e-05,
+      "loss": 0.0838,
+      "step": 5865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3495113282985342e-05,
+      "loss": 0.0975,
+      "step": 5870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3489560195468683e-05,
+      "loss": 0.0908,
+      "step": 5875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3484007107952022e-05,
+      "loss": 0.0591,
+      "step": 5880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3478454020435363e-05,
+      "loss": 0.058,
+      "step": 5885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3472900932918704e-05,
+      "loss": 0.1014,
+      "step": 5890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3467347845402045e-05,
+      "loss": 0.1578,
+      "step": 5895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3461794757885385e-05,
+      "loss": 0.2478,
+      "step": 5900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3456241670368726e-05,
+      "loss": 0.0728,
+      "step": 5905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3450688582852067e-05,
+      "loss": 0.109,
+      "step": 5910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3445135495335408e-05,
+      "loss": 0.091,
+      "step": 5915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3439582407818747e-05,
+      "loss": 0.1689,
+      "step": 5920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3434029320302088e-05,
+      "loss": 0.0528,
+      "step": 5925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.342847623278543e-05,
+      "loss": 0.0367,
+      "step": 5930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.342292314526877e-05,
+      "loss": 0.1093,
+      "step": 5935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.341737005775211e-05,
+      "loss": 0.1172,
+      "step": 5940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3411816970235451e-05,
+      "loss": 0.2165,
+      "step": 5945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3406263882718794e-05,
+      "loss": 0.3271,
+      "step": 5950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3400710795202135e-05,
+      "loss": 0.0653,
+      "step": 5955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3395157707685473e-05,
+      "loss": 0.0901,
+      "step": 5960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3389604620168815e-05,
+      "loss": 0.0854,
+      "step": 5965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3384051532652156e-05,
+      "loss": 0.0843,
+      "step": 5970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3378498445135497e-05,
+      "loss": 0.0888,
+      "step": 5975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3372945357618837e-05,
+      "loss": 0.0648,
+      "step": 5980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3367392270102178e-05,
+      "loss": 0.0728,
+      "step": 5985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3361839182585519e-05,
+      "loss": 0.1382,
+      "step": 5990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.335628609506886e-05,
+      "loss": 0.2024,
+      "step": 5995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.33507330075522e-05,
+      "loss": 0.3014,
+      "step": 6000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.036844413727521896,
+      "eval_runtime": 1820.2224,
+      "eval_samples_per_second": 8.8,
+      "eval_steps_per_second": 2.2,
+      "eval_wer": 0.1006130544536603,
+      "step": 6000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.334517992003554e-05,
+      "loss": 0.1017,
+      "step": 6005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3339626832518881e-05,
+      "loss": 0.096,
+      "step": 6010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3334073745002223e-05,
+      "loss": 0.0874,
+      "step": 6015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3328520657485564e-05,
+      "loss": 0.0932,
+      "step": 6020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3322967569968903e-05,
+      "loss": 0.0752,
+      "step": 6025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3317414482452244e-05,
+      "loss": 0.0853,
+      "step": 6030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3311861394935585e-05,
+      "loss": 0.0611,
+      "step": 6035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3306308307418928e-05,
+      "loss": 0.1506,
+      "step": 6040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3300755219902265e-05,
+      "loss": 0.2175,
+      "step": 6045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3295202132385607e-05,
+      "loss": 0.2922,
+      "step": 6050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.328964904486895e-05,
+      "loss": 0.0874,
+      "step": 6055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.328409595735229e-05,
+      "loss": 0.0906,
+      "step": 6060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.327854286983563e-05,
+      "loss": 0.0914,
+      "step": 6065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.327298978231897e-05,
+      "loss": 0.0975,
+      "step": 6070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3267436694802312e-05,
+      "loss": 0.0701,
+      "step": 6075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3261883607285653e-05,
+      "loss": 0.0423,
+      "step": 6080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3256330519768992e-05,
+      "loss": 0.0776,
+      "step": 6085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3250777432252333e-05,
+      "loss": 0.1417,
+      "step": 6090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3245224344735674e-05,
+      "loss": 0.1913,
+      "step": 6095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3239671257219015e-05,
+      "loss": 0.3044,
+      "step": 6100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3234118169702355e-05,
+      "loss": 0.0875,
+      "step": 6105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3228565082185696e-05,
+      "loss": 0.0929,
+      "step": 6110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3223011994669037e-05,
+      "loss": 0.0918,
+      "step": 6115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3217458907152378e-05,
+      "loss": 0.1143,
+      "step": 6120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3211905819635717e-05,
+      "loss": 0.0624,
+      "step": 6125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3206352732119058e-05,
+      "loss": 0.0334,
+      "step": 6130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.32007996446024e-05,
+      "loss": 0.1391,
+      "step": 6135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.319524655708574e-05,
+      "loss": 0.1837,
+      "step": 6140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3189693469569083e-05,
+      "loss": 0.1765,
+      "step": 6145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3184140382052421e-05,
+      "loss": 0.3795,
+      "step": 6150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3178587294535762e-05,
+      "loss": 0.0743,
+      "step": 6155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3173034207019105e-05,
+      "loss": 0.0849,
+      "step": 6160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3167481119502446e-05,
+      "loss": 0.0883,
+      "step": 6165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3161928031985785e-05,
+      "loss": 0.0848,
+      "step": 6170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3156374944469126e-05,
+      "loss": 0.0605,
+      "step": 6175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3150821856952467e-05,
+      "loss": 0.0624,
+      "step": 6180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3145268769435808e-05,
+      "loss": 0.0749,
+      "step": 6185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3139715681919148e-05,
+      "loss": 0.0968,
+      "step": 6190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3134162594402489e-05,
+      "loss": 0.1473,
+      "step": 6195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.312860950688583e-05,
+      "loss": 0.3527,
+      "step": 6200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3123056419369171e-05,
+      "loss": 0.1292,
+      "step": 6205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.311750333185251e-05,
+      "loss": 0.0994,
+      "step": 6210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3111950244335851e-05,
+      "loss": 0.0908,
+      "step": 6215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3106397156819193e-05,
+      "loss": 0.0855,
+      "step": 6220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3100844069302534e-05,
+      "loss": 0.064,
+      "step": 6225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3095290981785873e-05,
+      "loss": 0.0715,
+      "step": 6230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3089737894269214e-05,
+      "loss": 0.0828,
+      "step": 6235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3084184806752555e-05,
+      "loss": 0.0903,
+      "step": 6240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3078631719235896e-05,
+      "loss": 0.1712,
+      "step": 6245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3073078631719235e-05,
+      "loss": 0.2606,
+      "step": 6250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3067525544202577e-05,
+      "loss": 0.1096,
+      "step": 6255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.306197245668592e-05,
+      "loss": 0.0771,
+      "step": 6260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.305641936916926e-05,
+      "loss": 0.0835,
+      "step": 6265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3050866281652598e-05,
+      "loss": 0.1009,
+      "step": 6270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.304531319413594e-05,
+      "loss": 0.0881,
+      "step": 6275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3039760106619282e-05,
+      "loss": 0.0653,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3034207019102623e-05,
+      "loss": 0.0751,
+      "step": 6285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3028653931585964e-05,
+      "loss": 0.1077,
+      "step": 6290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3023100844069303e-05,
+      "loss": 0.1567,
+      "step": 6295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3017547756552644e-05,
+      "loss": 0.1985,
+      "step": 6300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3011994669035985e-05,
+      "loss": 0.1284,
+      "step": 6305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3006441581519327e-05,
+      "loss": 0.0916,
+      "step": 6310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3000888494002666e-05,
+      "loss": 0.0974,
+      "step": 6315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2995335406486007e-05,
+      "loss": 0.0967,
+      "step": 6320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2989782318969348e-05,
+      "loss": 0.0765,
+      "step": 6325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2984229231452689e-05,
+      "loss": 0.0407,
+      "step": 6330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2978676143936028e-05,
+      "loss": 0.046,
+      "step": 6335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.297312305641937e-05,
+      "loss": 0.0735,
+      "step": 6340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.296756996890271e-05,
+      "loss": 0.1319,
+      "step": 6345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2962016881386053e-05,
+      "loss": 0.276,
+      "step": 6350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2956463793869391e-05,
+      "loss": 0.0673,
+      "step": 6355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2950910706352732e-05,
+      "loss": 0.0854,
+      "step": 6360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2945357618836075e-05,
+      "loss": 0.0965,
+      "step": 6365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2939804531319416e-05,
+      "loss": 0.0917,
+      "step": 6370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2934251443802755e-05,
+      "loss": 0.0519,
+      "step": 6375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2928698356286096e-05,
+      "loss": 0.0543,
+      "step": 6380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2923145268769437e-05,
+      "loss": 0.1,
+      "step": 6385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2917592181252778e-05,
+      "loss": 0.1241,
+      "step": 6390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2912039093736118e-05,
+      "loss": 0.2167,
+      "step": 6395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2906486006219459e-05,
+      "loss": 0.3093,
+      "step": 6400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.29009329187028e-05,
+      "loss": 0.0783,
+      "step": 6405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2895379831186141e-05,
+      "loss": 0.0664,
+      "step": 6410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2889826743669482e-05,
+      "loss": 0.0665,
+      "step": 6415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2884273656152821e-05,
+      "loss": 0.0978,
+      "step": 6420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2878720568636163e-05,
+      "loss": 0.0608,
+      "step": 6425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2873167481119504e-05,
+      "loss": 0.0311,
+      "step": 6430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2867614393602845e-05,
+      "loss": 0.0914,
+      "step": 6435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2862061306086184e-05,
+      "loss": 0.1529,
+      "step": 6440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2856508218569525e-05,
+      "loss": 0.1729,
+      "step": 6445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2850955131052866e-05,
+      "loss": 0.2989,
+      "step": 6450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2845402043536209e-05,
+      "loss": 0.0629,
+      "step": 6455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2839848956019547e-05,
+      "loss": 0.0793,
+      "step": 6460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2834295868502888e-05,
+      "loss": 0.0734,
+      "step": 6465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.282874278098623e-05,
+      "loss": 0.118,
+      "step": 6470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2823189693469571e-05,
+      "loss": 0.0555,
+      "step": 6475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.281763660595291e-05,
+      "loss": 0.0655,
+      "step": 6480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2812083518436252e-05,
+      "loss": 0.094,
+      "step": 6485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2806530430919593e-05,
+      "loss": 0.0723,
+      "step": 6490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2800977343402934e-05,
+      "loss": 0.1638,
+      "step": 6495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2795424255886273e-05,
+      "loss": 0.3106,
+      "step": 6500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2789871168369614e-05,
+      "loss": 0.0656,
+      "step": 6505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2784318080852955e-05,
+      "loss": 0.0695,
+      "step": 6510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2778764993336297e-05,
+      "loss": 0.0917,
+      "step": 6515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2773211905819636e-05,
+      "loss": 0.1247,
+      "step": 6520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2767658818302977e-05,
+      "loss": 0.0545,
+      "step": 6525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2762105730786318e-05,
+      "loss": 0.0326,
+      "step": 6530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2756552643269659e-05,
+      "loss": 0.0536,
+      "step": 6535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2750999555752998e-05,
+      "loss": 0.0613,
+      "step": 6540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.274544646823634e-05,
+      "loss": 0.1258,
+      "step": 6545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.273989338071968e-05,
+      "loss": 0.266,
+      "step": 6550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2734340293203022e-05,
+      "loss": 0.0774,
+      "step": 6555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2728787205686364e-05,
+      "loss": 0.0719,
+      "step": 6560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2723234118169702e-05,
+      "loss": 0.0925,
+      "step": 6565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2717681030653045e-05,
+      "loss": 0.1228,
+      "step": 6570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2712127943136386e-05,
+      "loss": 0.1007,
+      "step": 6575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2706574855619727e-05,
+      "loss": 0.0623,
+      "step": 6580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2701021768103066e-05,
+      "loss": 0.0679,
+      "step": 6585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2695468680586407e-05,
+      "loss": 0.1125,
+      "step": 6590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2689915593069748e-05,
+      "loss": 0.1594,
+      "step": 6595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.268436250555309e-05,
+      "loss": 0.3766,
+      "step": 6600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2678809418036429e-05,
+      "loss": 0.0595,
+      "step": 6605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.267325633051977e-05,
+      "loss": 0.085,
+      "step": 6610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2667703243003111e-05,
+      "loss": 0.0736,
+      "step": 6615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2662150155486452e-05,
+      "loss": 0.0977,
+      "step": 6620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2656597067969791e-05,
+      "loss": 0.0707,
+      "step": 6625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2651043980453133e-05,
+      "loss": 0.0653,
+      "step": 6630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2645490892936474e-05,
+      "loss": 0.0562,
+      "step": 6635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2639937805419815e-05,
+      "loss": 0.1271,
+      "step": 6640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2634384717903154e-05,
+      "loss": 0.1889,
+      "step": 6645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2628831630386495e-05,
+      "loss": 0.2561,
+      "step": 6650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2623278542869836e-05,
+      "loss": 0.092,
+      "step": 6655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2617725455353179e-05,
+      "loss": 0.0901,
+      "step": 6660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2612172367836517e-05,
+      "loss": 0.0993,
+      "step": 6665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2606619280319858e-05,
+      "loss": 0.0945,
+      "step": 6670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.26010661928032e-05,
+      "loss": 0.0793,
+      "step": 6675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2595513105286541e-05,
+      "loss": 0.0381,
+      "step": 6680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2589960017769883e-05,
+      "loss": 0.1322,
+      "step": 6685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2584406930253222e-05,
+      "loss": 0.1044,
+      "step": 6690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2578853842736563e-05,
+      "loss": 0.1807,
+      "step": 6695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2573300755219904e-05,
+      "loss": 0.2764,
+      "step": 6700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2567747667703245e-05,
+      "loss": 0.0687,
+      "step": 6705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2562194580186584e-05,
+      "loss": 0.0821,
+      "step": 6710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2556641492669925e-05,
+      "loss": 0.0999,
+      "step": 6715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2551088405153267e-05,
+      "loss": 0.0868,
+      "step": 6720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2545535317636608e-05,
+      "loss": 0.0776,
+      "step": 6725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2539982230119947e-05,
+      "loss": 0.0365,
+      "step": 6730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2534429142603288e-05,
+      "loss": 0.0798,
+      "step": 6735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2528876055086629e-05,
+      "loss": 0.0935,
+      "step": 6740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.252332296756997e-05,
+      "loss": 0.1345,
+      "step": 6745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.251776988005331e-05,
+      "loss": 0.3742,
+      "step": 6750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.251221679253665e-05,
+      "loss": 0.0949,
+      "step": 6755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2506663705019992e-05,
+      "loss": 0.0604,
+      "step": 6760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2501110617503334e-05,
+      "loss": 0.0953,
+      "step": 6765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2495557529986672e-05,
+      "loss": 0.1049,
+      "step": 6770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2490004442470013e-05,
+      "loss": 0.0469,
+      "step": 6775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2484451354953356e-05,
+      "loss": 0.058,
+      "step": 6780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2478898267436697e-05,
+      "loss": 0.0696,
+      "step": 6785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2473345179920036e-05,
+      "loss": 0.1189,
+      "step": 6790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2467792092403377e-05,
+      "loss": 0.2272,
+      "step": 6795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2462239004886718e-05,
+      "loss": 0.3246,
+      "step": 6800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.245668591737006e-05,
+      "loss": 0.0933,
+      "step": 6805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2451132829853399e-05,
+      "loss": 0.0743,
+      "step": 6810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.244557974233674e-05,
+      "loss": 0.0858,
+      "step": 6815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2440026654820081e-05,
+      "loss": 0.1215,
+      "step": 6820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2434473567303422e-05,
+      "loss": 0.0811,
+      "step": 6825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2428920479786763e-05,
+      "loss": 0.0488,
+      "step": 6830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2423367392270103e-05,
+      "loss": 0.0944,
+      "step": 6835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2417814304753444e-05,
+      "loss": 0.0781,
+      "step": 6840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2412261217236785e-05,
+      "loss": 0.207,
+      "step": 6845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2406708129720126e-05,
+      "loss": 0.2736,
+      "step": 6850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2401155042203465e-05,
+      "loss": 0.0902,
+      "step": 6855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2395601954686806e-05,
+      "loss": 0.0724,
+      "step": 6860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2390048867170147e-05,
+      "loss": 0.0936,
+      "step": 6865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.238449577965349e-05,
+      "loss": 0.1051,
+      "step": 6870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2378942692136828e-05,
+      "loss": 0.0878,
+      "step": 6875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.237338960462017e-05,
+      "loss": 0.0677,
+      "step": 6880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2367836517103511e-05,
+      "loss": 0.0564,
+      "step": 6885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2362283429586853e-05,
+      "loss": 0.1026,
+      "step": 6890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2356730342070192e-05,
+      "loss": 0.1521,
+      "step": 6895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2351177254553533e-05,
+      "loss": 0.3306,
+      "step": 6900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2345624167036874e-05,
+      "loss": 0.0614,
+      "step": 6905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2340071079520215e-05,
+      "loss": 0.0764,
+      "step": 6910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2334517992003554e-05,
+      "loss": 0.076,
+      "step": 6915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2328964904486895e-05,
+      "loss": 0.0889,
+      "step": 6920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2323411816970237e-05,
+      "loss": 0.0642,
+      "step": 6925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2317858729453578e-05,
+      "loss": 0.0729,
+      "step": 6930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2312305641936917e-05,
+      "loss": 0.0646,
+      "step": 6935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2306752554420258e-05,
+      "loss": 0.123,
+      "step": 6940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2301199466903599e-05,
+      "loss": 0.1406,
+      "step": 6945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.229564637938694e-05,
+      "loss": 0.3075,
+      "step": 6950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.229009329187028e-05,
+      "loss": 0.0797,
+      "step": 6955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.228454020435362e-05,
+      "loss": 0.0726,
+      "step": 6960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2278987116836962e-05,
+      "loss": 0.0699,
+      "step": 6965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2273434029320304e-05,
+      "loss": 0.1097,
+      "step": 6970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2267880941803645e-05,
+      "loss": 0.068,
+      "step": 6975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2262327854286983e-05,
+      "loss": 0.0332,
+      "step": 6980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2256774766770326e-05,
+      "loss": 0.0682,
+      "step": 6985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2251221679253667e-05,
+      "loss": 0.1092,
+      "step": 6990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2245668591737008e-05,
+      "loss": 0.1286,
+      "step": 6995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2240115504220347e-05,
+      "loss": 0.2398,
+      "step": 7000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2234562416703688e-05,
+      "loss": 0.1131,
+      "step": 7005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.222900932918703e-05,
+      "loss": 0.0692,
+      "step": 7010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.222345624167037e-05,
+      "loss": 0.0974,
+      "step": 7015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.221790315415371e-05,
+      "loss": 0.0824,
+      "step": 7020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2212350066637051e-05,
+      "loss": 0.0727,
+      "step": 7025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2206796979120392e-05,
+      "loss": 0.0549,
+      "step": 7030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2201243891603733e-05,
+      "loss": 0.0778,
+      "step": 7035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2195690804087073e-05,
+      "loss": 0.1629,
+      "step": 7040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2190137716570414e-05,
+      "loss": 0.1378,
+      "step": 7045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2184584629053755e-05,
+      "loss": 0.2265,
+      "step": 7050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2179031541537096e-05,
+      "loss": 0.0768,
+      "step": 7055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2173478454020435e-05,
+      "loss": 0.0784,
+      "step": 7060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2167925366503776e-05,
+      "loss": 0.0943,
+      "step": 7065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2162372278987117e-05,
+      "loss": 0.1106,
+      "step": 7070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.215681919147046e-05,
+      "loss": 0.0887,
+      "step": 7075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2151266103953798e-05,
+      "loss": 0.0371,
+      "step": 7080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2145713016437139e-05,
+      "loss": 0.0688,
+      "step": 7085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2140159928920481e-05,
+      "loss": 0.1178,
+      "step": 7090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2134606841403823e-05,
+      "loss": 0.1882,
+      "step": 7095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2129053753887164e-05,
+      "loss": 0.2476,
+      "step": 7100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2123500666370503e-05,
+      "loss": 0.0826,
+      "step": 7105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2117947578853844e-05,
+      "loss": 0.0784,
+      "step": 7110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2112394491337185e-05,
+      "loss": 0.0658,
+      "step": 7115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2106841403820526e-05,
+      "loss": 0.0916,
+      "step": 7120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2101288316303865e-05,
+      "loss": 0.076,
+      "step": 7125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2095735228787207e-05,
+      "loss": 0.0296,
+      "step": 7130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2090182141270548e-05,
+      "loss": 0.0875,
+      "step": 7135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2084629053753889e-05,
+      "loss": 0.1144,
+      "step": 7140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2079075966237228e-05,
+      "loss": 0.1444,
+      "step": 7145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2073522878720569e-05,
+      "loss": 0.3095,
+      "step": 7150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.206796979120391e-05,
+      "loss": 0.0957,
+      "step": 7155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2062416703687251e-05,
+      "loss": 0.0835,
+      "step": 7160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.205686361617059e-05,
+      "loss": 0.0842,
+      "step": 7165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2051310528653932e-05,
+      "loss": 0.1096,
+      "step": 7170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2045757441137273e-05,
+      "loss": 0.0814,
+      "step": 7175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2040204353620615e-05,
+      "loss": 0.0317,
+      "step": 7180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2034651266103953e-05,
+      "loss": 0.0798,
+      "step": 7185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2029098178587296e-05,
+      "loss": 0.095,
+      "step": 7190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2023545091070637e-05,
+      "loss": 0.103,
+      "step": 7195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2017992003553978e-05,
+      "loss": 0.352,
+      "step": 7200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2012438916037317e-05,
+      "loss": 0.0673,
+      "step": 7205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2006885828520658e-05,
+      "loss": 0.0726,
+      "step": 7210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2001332741004e-05,
+      "loss": 0.0937,
+      "step": 7215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.199577965348734e-05,
+      "loss": 0.0993,
+      "step": 7220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.199022656597068e-05,
+      "loss": 0.1026,
+      "step": 7225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1984673478454021e-05,
+      "loss": 0.048,
+      "step": 7230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1979120390937362e-05,
+      "loss": 0.0614,
+      "step": 7235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1973567303420703e-05,
+      "loss": 0.0894,
+      "step": 7240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1968014215904044e-05,
+      "loss": 0.1637,
+      "step": 7245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1962461128387384e-05,
+      "loss": 0.2663,
+      "step": 7250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1956908040870725e-05,
+      "loss": 0.0688,
+      "step": 7255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1951354953354066e-05,
+      "loss": 0.0666,
+      "step": 7260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1945801865837407e-05,
+      "loss": 0.1073,
+      "step": 7265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1940248778320746e-05,
+      "loss": 0.0998,
+      "step": 7270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1934695690804087e-05,
+      "loss": 0.1142,
+      "step": 7275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.192914260328743e-05,
+      "loss": 0.0357,
+      "step": 7280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1923589515770771e-05,
+      "loss": 0.0909,
+      "step": 7285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1918036428254109e-05,
+      "loss": 0.076,
+      "step": 7290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1912483340737451e-05,
+      "loss": 0.1644,
+      "step": 7295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1906930253220793e-05,
+      "loss": 0.2738,
+      "step": 7300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1901377165704134e-05,
+      "loss": 0.0664,
+      "step": 7305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1895824078187473e-05,
+      "loss": 0.0726,
+      "step": 7310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1890270990670814e-05,
+      "loss": 0.1029,
+      "step": 7315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1884717903154155e-05,
+      "loss": 0.0958,
+      "step": 7320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1879164815637496e-05,
+      "loss": 0.0916,
+      "step": 7325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1873611728120835e-05,
+      "loss": 0.038,
+      "step": 7330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1868058640604177e-05,
+      "loss": 0.1326,
+      "step": 7335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1862505553087518e-05,
+      "loss": 0.1143,
+      "step": 7340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1856952465570859e-05,
+      "loss": 0.1256,
+      "step": 7345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1851399378054198e-05,
+      "loss": 0.2505,
+      "step": 7350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1845846290537539e-05,
+      "loss": 0.0835,
+      "step": 7355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.184029320302088e-05,
+      "loss": 0.0724,
+      "step": 7360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1834740115504221e-05,
+      "loss": 0.0806,
+      "step": 7365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1829187027987562e-05,
+      "loss": 0.103,
+      "step": 7370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1823633940470902e-05,
+      "loss": 0.0907,
+      "step": 7375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1818080852954243e-05,
+      "loss": 0.0558,
+      "step": 7380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1812527765437585e-05,
+      "loss": 0.0855,
+      "step": 7385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1806974677920927e-05,
+      "loss": 0.109,
+      "step": 7390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1801421590404264e-05,
+      "loss": 0.1699,
+      "step": 7395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1795868502887607e-05,
+      "loss": 0.1998,
+      "step": 7400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1790315415370948e-05,
+      "loss": 0.0815,
+      "step": 7405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1784762327854289e-05,
+      "loss": 0.0679,
+      "step": 7410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1779209240337628e-05,
+      "loss": 0.0747,
+      "step": 7415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.177365615282097e-05,
+      "loss": 0.1136,
+      "step": 7420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.176810306530431e-05,
+      "loss": 0.0891,
+      "step": 7425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1762549977787652e-05,
+      "loss": 0.0596,
+      "step": 7430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1756996890270991e-05,
+      "loss": 0.075,
+      "step": 7435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1751443802754332e-05,
+      "loss": 0.1402,
+      "step": 7440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1745890715237673e-05,
+      "loss": 0.1379,
+      "step": 7445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1740337627721014e-05,
+      "loss": 0.2628,
+      "step": 7450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1734784540204354e-05,
+      "loss": 0.0775,
+      "step": 7455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1729231452687695e-05,
+      "loss": 0.0763,
+      "step": 7460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1723678365171036e-05,
+      "loss": 0.09,
+      "step": 7465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1718125277654377e-05,
+      "loss": 0.1043,
+      "step": 7470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1712572190137716e-05,
+      "loss": 0.0497,
+      "step": 7475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1707019102621057e-05,
+      "loss": 0.0547,
+      "step": 7480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1701466015104398e-05,
+      "loss": 0.0468,
+      "step": 7485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1695912927587741e-05,
+      "loss": 0.0631,
+      "step": 7490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1690359840071079e-05,
+      "loss": 0.1546,
+      "step": 7495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1684806752554421e-05,
+      "loss": 0.3041,
+      "step": 7500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1679253665037763e-05,
+      "loss": 0.0867,
+      "step": 7505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1673700577521104e-05,
+      "loss": 0.0911,
+      "step": 7510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1668147490004445e-05,
+      "loss": 0.1018,
+      "step": 7515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1662594402487784e-05,
+      "loss": 0.0885,
+      "step": 7520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1657041314971125e-05,
+      "loss": 0.0657,
+      "step": 7525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1651488227454466e-05,
+      "loss": 0.0568,
+      "step": 7530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1645935139937807e-05,
+      "loss": 0.076,
+      "step": 7535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1640382052421147e-05,
+      "loss": 0.1338,
+      "step": 7540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1634828964904488e-05,
+      "loss": 0.1244,
+      "step": 7545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1629275877387829e-05,
+      "loss": 0.2539,
+      "step": 7550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.162372278987117e-05,
+      "loss": 0.0679,
+      "step": 7555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1618169702354509e-05,
+      "loss": 0.0776,
+      "step": 7560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.161261661483785e-05,
+      "loss": 0.0873,
+      "step": 7565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1607063527321191e-05,
+      "loss": 0.1074,
+      "step": 7570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1601510439804532e-05,
+      "loss": 0.0508,
+      "step": 7575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1595957352287872e-05,
+      "loss": 0.0702,
+      "step": 7580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1590404264771213e-05,
+      "loss": 0.0716,
+      "step": 7585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1584851177254555e-05,
+      "loss": 0.1019,
+      "step": 7590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1579298089737897e-05,
+      "loss": 0.1635,
+      "step": 7595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1573745002221234e-05,
+      "loss": 0.371,
+      "step": 7600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1568191914704577e-05,
+      "loss": 0.0837,
+      "step": 7605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1562638827187918e-05,
+      "loss": 0.0913,
+      "step": 7610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1557085739671259e-05,
+      "loss": 0.078,
+      "step": 7615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1551532652154598e-05,
+      "loss": 0.0948,
+      "step": 7620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154597956463794e-05,
+      "loss": 0.0505,
+      "step": 7625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154042647712128e-05,
+      "loss": 0.109,
+      "step": 7630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1534873389604622e-05,
+      "loss": 0.0851,
+      "step": 7635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1529320302087963e-05,
+      "loss": 0.0835,
+      "step": 7640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1523767214571302e-05,
+      "loss": 0.2377,
+      "step": 7645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518214127054643e-05,
+      "loss": 0.308,
+      "step": 7650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1512661039537984e-05,
+      "loss": 0.0853,
+      "step": 7655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507107952021325e-05,
+      "loss": 0.0764,
+      "step": 7660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1501554864504665e-05,
+      "loss": 0.1057,
+      "step": 7665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1496001776988006e-05,
+      "loss": 0.0947,
+      "step": 7670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1490448689471347e-05,
+      "loss": 0.0707,
+      "step": 7675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1484895601954688e-05,
+      "loss": 0.0622,
+      "step": 7680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1479342514438027e-05,
+      "loss": 0.0737,
+      "step": 7685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473789426921368e-05,
+      "loss": 0.1298,
+      "step": 7690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468236339404711e-05,
+      "loss": 0.1445,
+      "step": 7695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463793869391383e-05,
+      "loss": 0.5195,
+      "step": 7700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1458240781874724e-05,
+      "loss": 0.0784,
+      "step": 7705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1452687694358065e-05,
+      "loss": 0.08,
+      "step": 7710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1447134606841406e-05,
+      "loss": 0.112,
+      "step": 7715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1441581519324745e-05,
+      "loss": 0.0875,
+      "step": 7720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1436028431808086e-05,
+      "loss": 0.0986,
+      "step": 7725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1430475344291427e-05,
+      "loss": 0.0193,
+      "step": 7730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1424922256774768e-05,
+      "loss": 0.0789,
+      "step": 7735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1419369169258108e-05,
+      "loss": 0.0805,
+      "step": 7740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1413816081741449e-05,
+      "loss": 0.1398,
+      "step": 7745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.140826299422479e-05,
+      "loss": 0.33,
+      "step": 7750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.140270990670813e-05,
+      "loss": 0.0824,
+      "step": 7755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.139715681919147e-05,
+      "loss": 0.069,
+      "step": 7760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1391603731674811e-05,
+      "loss": 0.0972,
+      "step": 7765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1386050644158152e-05,
+      "loss": 0.1048,
+      "step": 7770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1380497556641495e-05,
+      "loss": 0.0684,
+      "step": 7775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1374944469124833e-05,
+      "loss": 0.0667,
+      "step": 7780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1369391381608175e-05,
+      "loss": 0.0953,
+      "step": 7785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1363838294091517e-05,
+      "loss": 0.1007,
+      "step": 7790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1358285206574858e-05,
+      "loss": 0.1114,
+      "step": 7795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1352732119058197e-05,
+      "loss": 0.3188,
+      "step": 7800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1347179031541538e-05,
+      "loss": 0.0714,
+      "step": 7805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1341625944024879e-05,
+      "loss": 0.0624,
+      "step": 7810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.133607285650822e-05,
+      "loss": 0.0733,
+      "step": 7815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.133051976899156e-05,
+      "loss": 0.1139,
+      "step": 7820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.13249666814749e-05,
+      "loss": 0.0523,
+      "step": 7825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1319413593958242e-05,
+      "loss": 0.0635,
+      "step": 7830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1313860506441583e-05,
+      "loss": 0.1009,
+      "step": 7835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1308307418924924e-05,
+      "loss": 0.0955,
+      "step": 7840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1302754331408263e-05,
+      "loss": 0.1388,
+      "step": 7845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1297201243891604e-05,
+      "loss": 0.2546,
+      "step": 7850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1291648156374945e-05,
+      "loss": 0.0709,
+      "step": 7855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1286095068858286e-05,
+      "loss": 0.0806,
+      "step": 7860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1280541981341626e-05,
+      "loss": 0.0654,
+      "step": 7865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1274988893824967e-05,
+      "loss": 0.0894,
+      "step": 7870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.126943580630831e-05,
+      "loss": 0.0977,
+      "step": 7875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.126388271879165e-05,
+      "loss": 0.0561,
+      "step": 7880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1258329631274988e-05,
+      "loss": 0.0614,
+      "step": 7885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1252776543758331e-05,
+      "loss": 0.111,
+      "step": 7890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1247223456241672e-05,
+      "loss": 0.1592,
+      "step": 7895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1241670368725013e-05,
+      "loss": 0.2755,
+      "step": 7900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1236117281208353e-05,
+      "loss": 0.0819,
+      "step": 7905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1230564193691694e-05,
+      "loss": 0.0998,
+      "step": 7910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1225011106175035e-05,
+      "loss": 0.0772,
+      "step": 7915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1219458018658376e-05,
+      "loss": 0.0856,
+      "step": 7920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1213904931141715e-05,
+      "loss": 0.0872,
+      "step": 7925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1208351843625056e-05,
+      "loss": 0.038,
+      "step": 7930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1202798756108397e-05,
+      "loss": 0.1082,
+      "step": 7935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1197245668591738e-05,
+      "loss": 0.156,
+      "step": 7940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1191692581075078e-05,
+      "loss": 0.1274,
+      "step": 7945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1186139493558419e-05,
+      "loss": 0.2681,
+      "step": 7950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.118058640604176e-05,
+      "loss": 0.0749,
+      "step": 7955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.11750333185251e-05,
+      "loss": 0.0729,
+      "step": 7960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1169480231008442e-05,
+      "loss": 0.0877,
+      "step": 7965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1163927143491781e-05,
+      "loss": 0.0814,
+      "step": 7970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1158374055975122e-05,
+      "loss": 0.0642,
+      "step": 7975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1152820968458465e-05,
+      "loss": 0.0322,
+      "step": 7980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1147267880941806e-05,
+      "loss": 0.0572,
+      "step": 7985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1141714793425144e-05,
+      "loss": 0.104,
+      "step": 7990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1136161705908487e-05,
+      "loss": 0.111,
+      "step": 7995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1130608618391828e-05,
+      "loss": 0.302,
+      "step": 8000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.03474867343902588,
+      "eval_runtime": 1810.8597,
+      "eval_samples_per_second": 8.846,
+      "eval_steps_per_second": 2.212,
+      "eval_wer": 0.08851424450054093,
+      "step": 8000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1125055530875169e-05,
+      "loss": 0.106,
+      "step": 8005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1119502443358508e-05,
+      "loss": 0.0708,
+      "step": 8010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1113949355841849e-05,
+      "loss": 0.1007,
+      "step": 8015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.110839626832519e-05,
+      "loss": 0.0792,
+      "step": 8020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1102843180808531e-05,
+      "loss": 0.1007,
+      "step": 8025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.109729009329187e-05,
+      "loss": 0.0455,
+      "step": 8030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1091737005775212e-05,
+      "loss": 0.0634,
+      "step": 8035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1086183918258553e-05,
+      "loss": 0.0717,
+      "step": 8040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1080630830741894e-05,
+      "loss": 0.1224,
+      "step": 8045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1075077743225233e-05,
+      "loss": 0.3224,
+      "step": 8050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1069524655708574e-05,
+      "loss": 0.0863,
+      "step": 8055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1063971568191915e-05,
+      "loss": 0.0761,
+      "step": 8060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1058418480675256e-05,
+      "loss": 0.1265,
+      "step": 8065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1052865393158596e-05,
+      "loss": 0.1053,
+      "step": 8070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1047312305641937e-05,
+      "loss": 0.0667,
+      "step": 8075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1041759218125278e-05,
+      "loss": 0.0568,
+      "step": 8080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.103620613060862e-05,
+      "loss": 0.0537,
+      "step": 8085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1030653043091958e-05,
+      "loss": 0.1009,
+      "step": 8090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1025099955575301e-05,
+      "loss": 0.1139,
+      "step": 8095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1019546868058642e-05,
+      "loss": 0.2392,
+      "step": 8100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1013993780541983e-05,
+      "loss": 0.091,
+      "step": 8105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1008440693025324e-05,
+      "loss": 0.0759,
+      "step": 8110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1002887605508664e-05,
+      "loss": 0.0879,
+      "step": 8115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0997334517992005e-05,
+      "loss": 0.09,
+      "step": 8120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0991781430475346e-05,
+      "loss": 0.0958,
+      "step": 8125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0986228342958687e-05,
+      "loss": 0.0568,
+      "step": 8130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0980675255442026e-05,
+      "loss": 0.1016,
+      "step": 8135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0975122167925367e-05,
+      "loss": 0.1496,
+      "step": 8140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0969569080408708e-05,
+      "loss": 0.1523,
+      "step": 8145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.096401599289205e-05,
+      "loss": 0.2466,
+      "step": 8150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0958462905375389e-05,
+      "loss": 0.062,
+      "step": 8155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.095290981785873e-05,
+      "loss": 0.0759,
+      "step": 8160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.094735673034207e-05,
+      "loss": 0.0717,
+      "step": 8165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0941803642825412e-05,
+      "loss": 0.1179,
+      "step": 8170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0936250555308751e-05,
+      "loss": 0.0686,
+      "step": 8175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0930697467792092e-05,
+      "loss": 0.031,
+      "step": 8180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0925144380275435e-05,
+      "loss": 0.0653,
+      "step": 8185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0919591292758776e-05,
+      "loss": 0.1195,
+      "step": 8190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0914038205242114e-05,
+      "loss": 0.2072,
+      "step": 8195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0908485117725457e-05,
+      "loss": 0.2945,
+      "step": 8200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0902932030208798e-05,
+      "loss": 0.0873,
+      "step": 8205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0897378942692139e-05,
+      "loss": 0.0985,
+      "step": 8210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0891825855175478e-05,
+      "loss": 0.0986,
+      "step": 8215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0886272767658819e-05,
+      "loss": 0.1102,
+      "step": 8220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.088071968014216e-05,
+      "loss": 0.075,
+      "step": 8225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0875166592625501e-05,
+      "loss": 0.0354,
+      "step": 8230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0869613505108842e-05,
+      "loss": 0.0868,
+      "step": 8235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0864060417592182e-05,
+      "loss": 0.1088,
+      "step": 8240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0858507330075523e-05,
+      "loss": 0.1612,
+      "step": 8245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0852954242558864e-05,
+      "loss": 0.235,
+      "step": 8250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0847401155042205e-05,
+      "loss": 0.0823,
+      "step": 8255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0841848067525544e-05,
+      "loss": 0.0721,
+      "step": 8260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0836294980008885e-05,
+      "loss": 0.0969,
+      "step": 8265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0830741892492226e-05,
+      "loss": 0.1035,
+      "step": 8270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0825188804975567e-05,
+      "loss": 0.0786,
+      "step": 8275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0819635717458907e-05,
+      "loss": 0.0562,
+      "step": 8280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0814082629942248e-05,
+      "loss": 0.0721,
+      "step": 8285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.080852954242559e-05,
+      "loss": 0.1153,
+      "step": 8290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0802976454908932e-05,
+      "loss": 0.2193,
+      "step": 8295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.079742336739227e-05,
+      "loss": 0.2758,
+      "step": 8300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0791870279875612e-05,
+      "loss": 0.0604,
+      "step": 8305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0786317192358953e-05,
+      "loss": 0.0817,
+      "step": 8310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0780764104842294e-05,
+      "loss": 0.0737,
+      "step": 8315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0775211017325634e-05,
+      "loss": 0.0984,
+      "step": 8320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0769657929808975e-05,
+      "loss": 0.0749,
+      "step": 8325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0764104842292316e-05,
+      "loss": 0.0421,
+      "step": 8330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0758551754775657e-05,
+      "loss": 0.0684,
+      "step": 8335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0752998667258996e-05,
+      "loss": 0.0969,
+      "step": 8340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0747445579742337e-05,
+      "loss": 0.1404,
+      "step": 8345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0741892492225678e-05,
+      "loss": 0.2184,
+      "step": 8350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.073633940470902e-05,
+      "loss": 0.075,
+      "step": 8355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0730786317192359e-05,
+      "loss": 0.083,
+      "step": 8360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.07252332296757e-05,
+      "loss": 0.0745,
+      "step": 8365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.071968014215904e-05,
+      "loss": 0.0828,
+      "step": 8370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0714127054642382e-05,
+      "loss": 0.0698,
+      "step": 8375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0708573967125725e-05,
+      "loss": 0.0825,
+      "step": 8380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0703020879609062e-05,
+      "loss": 0.0779,
+      "step": 8385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0697467792092403e-05,
+      "loss": 0.1404,
+      "step": 8390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0691914704575746e-05,
+      "loss": 0.1043,
+      "step": 8395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0686361617059087e-05,
+      "loss": 0.2751,
+      "step": 8400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0680808529542427e-05,
+      "loss": 0.094,
+      "step": 8405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0675255442025768e-05,
+      "loss": 0.0883,
+      "step": 8410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0669702354509109e-05,
+      "loss": 0.1024,
+      "step": 8415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.066414926699245e-05,
+      "loss": 0.0884,
+      "step": 8420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0658596179475789e-05,
+      "loss": 0.0859,
+      "step": 8425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.065304309195913e-05,
+      "loss": 0.0794,
+      "step": 8430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0647490004442471e-05,
+      "loss": 0.0772,
+      "step": 8435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0641936916925812e-05,
+      "loss": 0.0931,
+      "step": 8440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0636383829409152e-05,
+      "loss": 0.1078,
+      "step": 8445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0630830741892493e-05,
+      "loss": 0.2906,
+      "step": 8450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0625277654375834e-05,
+      "loss": 0.0769,
+      "step": 8455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0619724566859175e-05,
+      "loss": 0.0995,
+      "step": 8460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0614171479342514e-05,
+      "loss": 0.1122,
+      "step": 8465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0608618391825855e-05,
+      "loss": 0.1084,
+      "step": 8470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0603065304309196e-05,
+      "loss": 0.0693,
+      "step": 8475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0597512216792537e-05,
+      "loss": 0.0603,
+      "step": 8480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0591959129275877e-05,
+      "loss": 0.0701,
+      "step": 8485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0586406041759218e-05,
+      "loss": 0.1333,
+      "step": 8490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.058085295424256e-05,
+      "loss": 0.1673,
+      "step": 8495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0575299866725902e-05,
+      "loss": 0.2957,
+      "step": 8500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0569746779209243e-05,
+      "loss": 0.0753,
+      "step": 8505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0564193691692582e-05,
+      "loss": 0.0823,
+      "step": 8510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0558640604175923e-05,
+      "loss": 0.081,
+      "step": 8515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0553087516659264e-05,
+      "loss": 0.108,
+      "step": 8520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0547534429142605e-05,
+      "loss": 0.1037,
+      "step": 8525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0541981341625945e-05,
+      "loss": 0.0428,
+      "step": 8530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0536428254109286e-05,
+      "loss": 0.0715,
+      "step": 8535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0530875166592627e-05,
+      "loss": 0.0539,
+      "step": 8540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0525322079075968e-05,
+      "loss": 0.1419,
+      "step": 8545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0519768991559307e-05,
+      "loss": 0.2654,
+      "step": 8550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0514215904042648e-05,
+      "loss": 0.0642,
+      "step": 8555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.050866281652599e-05,
+      "loss": 0.0741,
+      "step": 8560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.050310972900933e-05,
+      "loss": 0.0991,
+      "step": 8565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.049755664149267e-05,
+      "loss": 0.0959,
+      "step": 8570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.049200355397601e-05,
+      "loss": 0.0606,
+      "step": 8575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0486450466459352e-05,
+      "loss": 0.0777,
+      "step": 8580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0480897378942693e-05,
+      "loss": 0.0993,
+      "step": 8585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0475344291426032e-05,
+      "loss": 0.0984,
+      "step": 8590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0469791203909373e-05,
+      "loss": 0.1397,
+      "step": 8595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0464238116392716e-05,
+      "loss": 0.2935,
+      "step": 8600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0458685028876057e-05,
+      "loss": 0.0781,
+      "step": 8605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0453131941359395e-05,
+      "loss": 0.0568,
+      "step": 8610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0447578853842738e-05,
+      "loss": 0.1006,
+      "step": 8615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0442025766326079e-05,
+      "loss": 0.0871,
+      "step": 8620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043647267880942e-05,
+      "loss": 0.0733,
+      "step": 8625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0430919591292759e-05,
+      "loss": 0.0508,
+      "step": 8630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.04253665037761e-05,
+      "loss": 0.0861,
+      "step": 8635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0419813416259441e-05,
+      "loss": 0.1221,
+      "step": 8640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0414260328742782e-05,
+      "loss": 0.1375,
+      "step": 8645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0408707241226123e-05,
+      "loss": 0.2757,
+      "step": 8650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0403154153709463e-05,
+      "loss": 0.072,
+      "step": 8655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0397601066192804e-05,
+      "loss": 0.0904,
+      "step": 8660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0392047978676145e-05,
+      "loss": 0.0962,
+      "step": 8665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0386494891159486e-05,
+      "loss": 0.054,
+      "step": 8670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0380941803642825e-05,
+      "loss": 0.0826,
+      "step": 8675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0375388716126166e-05,
+      "loss": 0.0445,
+      "step": 8680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0369835628609507e-05,
+      "loss": 0.0709,
+      "step": 8685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.036428254109285e-05,
+      "loss": 0.0767,
+      "step": 8690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0358729453576188e-05,
+      "loss": 0.1863,
+      "step": 8695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0353176366059529e-05,
+      "loss": 0.3167,
+      "step": 8700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0347623278542872e-05,
+      "loss": 0.0604,
+      "step": 8705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0342070191026213e-05,
+      "loss": 0.0765,
+      "step": 8710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0336517103509552e-05,
+      "loss": 0.0985,
+      "step": 8715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0330964015992893e-05,
+      "loss": 0.1063,
+      "step": 8720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0325410928476234e-05,
+      "loss": 0.0517,
+      "step": 8725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0319857840959575e-05,
+      "loss": 0.0216,
+      "step": 8730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0314304753442915e-05,
+      "loss": 0.0608,
+      "step": 8735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0308751665926256e-05,
+      "loss": 0.1446,
+      "step": 8740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0303198578409597e-05,
+      "loss": 0.1741,
+      "step": 8745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0297645490892938e-05,
+      "loss": 0.193,
+      "step": 8750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0292092403376277e-05,
+      "loss": 0.0618,
+      "step": 8755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0286539315859618e-05,
+      "loss": 0.0628,
+      "step": 8760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.028098622834296e-05,
+      "loss": 0.1094,
+      "step": 8765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.02754331408263e-05,
+      "loss": 0.0873,
+      "step": 8770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0269880053309641e-05,
+      "loss": 0.0766,
+      "step": 8775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.026432696579298e-05,
+      "loss": 0.0489,
+      "step": 8780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0258773878276322e-05,
+      "loss": 0.0592,
+      "step": 8785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0253220790759663e-05,
+      "loss": 0.0832,
+      "step": 8790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0247667703243006e-05,
+      "loss": 0.1182,
+      "step": 8795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0242114615726343e-05,
+      "loss": 0.3355,
+      "step": 8800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0236561528209686e-05,
+      "loss": 0.0623,
+      "step": 8805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0231008440693027e-05,
+      "loss": 0.0657,
+      "step": 8810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0225455353176368e-05,
+      "loss": 0.0858,
+      "step": 8815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0219902265659708e-05,
+      "loss": 0.0962,
+      "step": 8820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0214349178143049e-05,
+      "loss": 0.0589,
+      "step": 8825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.020879609062639e-05,
+      "loss": 0.1494,
+      "step": 8830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.020324300310973e-05,
+      "loss": 0.0902,
+      "step": 8835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.019768991559307e-05,
+      "loss": 0.1051,
+      "step": 8840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0192136828076411e-05,
+      "loss": 0.1033,
+      "step": 8845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0186583740559752e-05,
+      "loss": 0.3234,
+      "step": 8850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0181030653043093e-05,
+      "loss": 0.0701,
+      "step": 8855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0175477565526433e-05,
+      "loss": 0.0662,
+      "step": 8860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0169924478009774e-05,
+      "loss": 0.0727,
+      "step": 8865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0164371390493115e-05,
+      "loss": 0.1058,
+      "step": 8870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0158818302976456e-05,
+      "loss": 0.0291,
+      "step": 8875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0153265215459795e-05,
+      "loss": 0.0796,
+      "step": 8880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0147712127943136e-05,
+      "loss": 0.1045,
+      "step": 8885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0142159040426477e-05,
+      "loss": 0.1044,
+      "step": 8890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0136605952909818e-05,
+      "loss": 0.189,
+      "step": 8895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0131052865393158e-05,
+      "loss": 0.2141,
+      "step": 8900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0125499777876499e-05,
+      "loss": 0.0881,
+      "step": 8905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0119946690359842e-05,
+      "loss": 0.0875,
+      "step": 8910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0114393602843183e-05,
+      "loss": 0.0863,
+      "step": 8915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0108840515326524e-05,
+      "loss": 0.089,
+      "step": 8920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0103287427809863e-05,
+      "loss": 0.0462,
+      "step": 8925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0097734340293204e-05,
+      "loss": 0.0526,
+      "step": 8930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0092181252776545e-05,
+      "loss": 0.0568,
+      "step": 8935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0086628165259886e-05,
+      "loss": 0.1226,
+      "step": 8940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0081075077743226e-05,
+      "loss": 0.2144,
+      "step": 8945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0075521990226567e-05,
+      "loss": 0.3195,
+      "step": 8950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0069968902709908e-05,
+      "loss": 0.0534,
+      "step": 8955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0064415815193249e-05,
+      "loss": 0.0817,
+      "step": 8960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0058862727676588e-05,
+      "loss": 0.0665,
+      "step": 8965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.005330964015993e-05,
+      "loss": 0.0908,
+      "step": 8970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.004775655264327e-05,
+      "loss": 0.0749,
+      "step": 8975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0042203465126611e-05,
+      "loss": 0.044,
+      "step": 8980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.003665037760995e-05,
+      "loss": 0.0613,
+      "step": 8985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0031097290093292e-05,
+      "loss": 0.1023,
+      "step": 8990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0025544202576633e-05,
+      "loss": 0.143,
+      "step": 8995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0019991115059976e-05,
+      "loss": 0.2331,
+      "step": 9000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0014438027543313e-05,
+      "loss": 0.0839,
+      "step": 9005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0008884940026654e-05,
+      "loss": 0.1403,
+      "step": 9010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0003331852509997e-05,
+      "loss": 0.0757,
+      "step": 9015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.997778764993337e-06,
+      "loss": 0.093,
+      "step": 9020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.992225677476678e-06,
+      "loss": 0.0854,
+      "step": 9025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.986672589960019e-06,
+      "loss": 0.084,
+      "step": 9030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.98111950244336e-06,
+      "loss": 0.0606,
+      "step": 9035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.975566414926699e-06,
+      "loss": 0.051,
+      "step": 9040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.97001332741004e-06,
+      "loss": 0.0723,
+      "step": 9045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.964460239893381e-06,
+      "loss": 0.1134,
+      "step": 9050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.958907152376722e-06,
+      "loss": 0.2029,
+      "step": 9055
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.953354064860063e-06,
+      "loss": 0.2837,
+      "step": 9060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.947800977343404e-06,
+      "loss": 0.0507,
+      "step": 9065
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.942247889826745e-06,
+      "loss": 0.0734,
+      "step": 9070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.936694802310085e-06,
+      "loss": 0.0853,
+      "step": 9075
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.931141714793426e-06,
+      "loss": 0.0952,
+      "step": 9080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.925588627276767e-06,
+      "loss": 0.0532,
+      "step": 9085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.920035539760108e-06,
+      "loss": 0.0337,
+      "step": 9090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.914482452243447e-06,
+      "loss": 0.0653,
+      "step": 9095
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.908929364726788e-06,
+      "loss": 0.0994,
+      "step": 9100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.90337627721013e-06,
+      "loss": 0.0857,
+      "step": 9105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.89782318969347e-06,
+      "loss": 0.2042,
+      "step": 9110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.892270102176812e-06,
+      "loss": 0.07,
+      "step": 9115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.886717014660153e-06,
+      "loss": 0.0933,
+      "step": 9120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.881163927143492e-06,
+      "loss": 0.063,
+      "step": 9125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.875610839626833e-06,
+      "loss": 0.0854,
+      "step": 9130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.870057752110174e-06,
+      "loss": 0.0575,
+      "step": 9135
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.864504664593515e-06,
+      "loss": 0.0519,
+      "step": 9140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.858951577076855e-06,
+      "loss": 0.0573,
+      "step": 9145
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.853398489560196e-06,
+      "loss": 0.1191,
+      "step": 9150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.847845402043537e-06,
+      "loss": 0.1459,
+      "step": 9155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.842292314526878e-06,
+      "loss": 0.2295,
+      "step": 9160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.836739227010217e-06,
+      "loss": 0.0642,
+      "step": 9165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.83118613949356e-06,
+      "loss": 0.0675,
+      "step": 9170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.8256330519769e-06,
+      "loss": 0.0686,
+      "step": 9175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.82007996446024e-06,
+      "loss": 0.1142,
+      "step": 9180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.814526876943581e-06,
+      "loss": 0.0696,
+      "step": 9185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.808973789426923e-06,
+      "loss": 0.0416,
+      "step": 9190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.803420701910264e-06,
+      "loss": 0.0972,
+      "step": 9195
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.797867614393603e-06,
+      "loss": 0.1283,
+      "step": 9200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.792314526876944e-06,
+      "loss": 0.1552,
+      "step": 9205
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.786761439360285e-06,
+      "loss": 0.2492,
+      "step": 9210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.781208351843626e-06,
+      "loss": 0.0781,
+      "step": 9215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.775655264326967e-06,
+      "loss": 0.0934,
+      "step": 9220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.770102176810308e-06,
+      "loss": 0.0825,
+      "step": 9225
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.764549089293648e-06,
+      "loss": 0.1004,
+      "step": 9230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.758996001776989e-06,
+      "loss": 0.0455,
+      "step": 9235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.75344291426033e-06,
+      "loss": 0.084,
+      "step": 9240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.74788982674367e-06,
+      "loss": 0.0802,
+      "step": 9245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.74233673922701e-06,
+      "loss": 0.1308,
+      "step": 9250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.736783651710351e-06,
+      "loss": 0.1009,
+      "step": 9255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.731230564193692e-06,
+      "loss": 0.2178,
+      "step": 9260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.725677476677033e-06,
+      "loss": 0.0614,
+      "step": 9265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.720124389160374e-06,
+      "loss": 0.1016,
+      "step": 9270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.714571301643715e-06,
+      "loss": 0.0839,
+      "step": 9275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.709018214127055e-06,
+      "loss": 0.0835,
+      "step": 9280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.703465126610396e-06,
+      "loss": 0.0296,
+      "step": 9285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.697912039093737e-06,
+      "loss": 0.0332,
+      "step": 9290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.692358951577078e-06,
+      "loss": 0.0958,
+      "step": 9295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.686805864060417e-06,
+      "loss": 0.1346,
+      "step": 9300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.681252776543758e-06,
+      "loss": 0.1797,
+      "step": 9305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.6756996890271e-06,
+      "loss": 0.2794,
+      "step": 9310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.67014660151044e-06,
+      "loss": 0.0839,
+      "step": 9315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.664593513993782e-06,
+      "loss": 0.0717,
+      "step": 9320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.659040426477123e-06,
+      "loss": 0.0909,
+      "step": 9325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.653487338960464e-06,
+      "loss": 0.0772,
+      "step": 9330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.647934251443803e-06,
+      "loss": 0.0809,
+      "step": 9335
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.642381163927144e-06,
+      "loss": 0.0698,
+      "step": 9340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.636828076410485e-06,
+      "loss": 0.0991,
+      "step": 9345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.631274988893826e-06,
+      "loss": 0.1132,
+      "step": 9350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.625721901377166e-06,
+      "loss": 0.1813,
+      "step": 9355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.620168813860507e-06,
+      "loss": 0.2971,
+      "step": 9360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.614615726343848e-06,
+      "loss": 0.0776,
+      "step": 9365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.609062638827189e-06,
+      "loss": 0.0964,
+      "step": 9370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.60350955131053e-06,
+      "loss": 0.0816,
+      "step": 9375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.597956463793871e-06,
+      "loss": 0.1118,
+      "step": 9380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.59240337627721e-06,
+      "loss": 0.0642,
+      "step": 9385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.586850288760551e-06,
+      "loss": 0.0904,
+      "step": 9390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.581297201243893e-06,
+      "loss": 0.1273,
+      "step": 9395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.575744113727234e-06,
+      "loss": 0.1431,
+      "step": 9400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.570191026210573e-06,
+      "loss": 0.1998,
+      "step": 9405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.564637938693914e-06,
+      "loss": 0.2178,
+      "step": 9410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.559084851177255e-06,
+      "loss": 0.0652,
+      "step": 9415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.553531763660596e-06,
+      "loss": 0.0782,
+      "step": 9420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.547978676143937e-06,
+      "loss": 0.0775,
+      "step": 9425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.542425588627278e-06,
+      "loss": 0.0878,
+      "step": 9430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.536872501110618e-06,
+      "loss": 0.0363,
+      "step": 9435
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.531319413593959e-06,
+      "loss": 0.0745,
+      "step": 9440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.5257663260773e-06,
+      "loss": 0.0608,
+      "step": 9445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.52021323856064e-06,
+      "loss": 0.1054,
+      "step": 9450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.514660151043982e-06,
+      "loss": 0.1699,
+      "step": 9455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.509107063527321e-06,
+      "loss": 0.2034,
+      "step": 9460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.503553976010664e-06,
+      "loss": 0.0742,
+      "step": 9465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.498000888494003e-06,
+      "loss": 0.0641,
+      "step": 9470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.492447800977344e-06,
+      "loss": 0.0714,
+      "step": 9475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.486894713460685e-06,
+      "loss": 0.1139,
+      "step": 9480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.481341625944027e-06,
+      "loss": 0.0519,
+      "step": 9485
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.475788538427366e-06,
+      "loss": 0.058,
+      "step": 9490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.470235450910707e-06,
+      "loss": 0.0923,
+      "step": 9495
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.464682363394048e-06,
+      "loss": 0.115,
+      "step": 9500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.459129275877389e-06,
+      "loss": 0.2176,
+      "step": 9505
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.453576188360728e-06,
+      "loss": 0.2731,
+      "step": 9510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.44802310084407e-06,
+      "loss": 0.0734,
+      "step": 9515
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.44247001332741e-06,
+      "loss": 0.0653,
+      "step": 9520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.436916925810752e-06,
+      "loss": 0.0732,
+      "step": 9525
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.431363838294093e-06,
+      "loss": 0.0794,
+      "step": 9530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.425810750777434e-06,
+      "loss": 0.068,
+      "step": 9535
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.420257663260773e-06,
+      "loss": 0.0973,
+      "step": 9540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.414704575744114e-06,
+      "loss": 0.1599,
+      "step": 9545
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.409151488227455e-06,
+      "loss": 0.1473,
+      "step": 9550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.403598400710796e-06,
+      "loss": 0.1827,
+      "step": 9555
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.398045313194136e-06,
+      "loss": 0.2009,
+      "step": 9560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.392492225677477e-06,
+      "loss": 0.0616,
+      "step": 9565
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.386939138160818e-06,
+      "loss": 0.0808,
+      "step": 9570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.381386050644159e-06,
+      "loss": 0.0805,
+      "step": 9575
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.3758329631275e-06,
+      "loss": 0.0839,
+      "step": 9580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.370279875610841e-06,
+      "loss": 0.0395,
+      "step": 9585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.364726788094182e-06,
+      "loss": 0.0223,
+      "step": 9590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.359173700577521e-06,
+      "loss": 0.0583,
+      "step": 9595
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.353620613060863e-06,
+      "loss": 0.0823,
+      "step": 9600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.348067525544204e-06,
+      "loss": 0.138,
+      "step": 9605
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.342514438027545e-06,
+      "loss": 0.3078,
+      "step": 9610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.336961350510884e-06,
+      "loss": 0.0712,
+      "step": 9615
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.331408262994227e-06,
+      "loss": 0.0754,
+      "step": 9620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.325855175477566e-06,
+      "loss": 0.0889,
+      "step": 9625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.320302087960907e-06,
+      "loss": 0.1146,
+      "step": 9630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.314749000444248e-06,
+      "loss": 0.0488,
+      "step": 9635
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.30919591292759e-06,
+      "loss": 0.0445,
+      "step": 9640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.303642825410929e-06,
+      "loss": 0.0777,
+      "step": 9645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.29808973789427e-06,
+      "loss": 0.1293,
+      "step": 9650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.29253665037761e-06,
+      "loss": 0.1418,
+      "step": 9655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.286983562860952e-06,
+      "loss": 0.2381,
+      "step": 9660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.281430475344291e-06,
+      "loss": 0.0638,
+      "step": 9665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.275877387827632e-06,
+      "loss": 0.0684,
+      "step": 9670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.270324300310973e-06,
+      "loss": 0.0814,
+      "step": 9675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.264771212794314e-06,
+      "loss": 0.1062,
+      "step": 9680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.259218125277655e-06,
+      "loss": 0.0433,
+      "step": 9685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.253665037760997e-06,
+      "loss": 0.049,
+      "step": 9690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.248111950244336e-06,
+      "loss": 0.0646,
+      "step": 9695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.242558862727677e-06,
+      "loss": 0.1562,
+      "step": 9700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.237005775211018e-06,
+      "loss": 0.1716,
+      "step": 9705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.231452687694359e-06,
+      "loss": 0.2675,
+      "step": 9710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.225899600177698e-06,
+      "loss": 0.076,
+      "step": 9715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.22034651266104e-06,
+      "loss": 0.0608,
+      "step": 9720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.214793425144382e-06,
+      "loss": 0.0845,
+      "step": 9725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.209240337627722e-06,
+      "loss": 0.0688,
+      "step": 9730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.203687250111063e-06,
+      "loss": 0.0523,
+      "step": 9735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.198134162594404e-06,
+      "loss": 0.0277,
+      "step": 9740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.192581075077745e-06,
+      "loss": 0.0661,
+      "step": 9745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.187027987561084e-06,
+      "loss": 0.1164,
+      "step": 9750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.181474900044425e-06,
+      "loss": 0.111,
+      "step": 9755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.175921812527766e-06,
+      "loss": 0.2074,
+      "step": 9760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.170368725011107e-06,
+      "loss": 0.0627,
+      "step": 9765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.164815637494447e-06,
+      "loss": 0.132,
+      "step": 9770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.15926254997779e-06,
+      "loss": 0.0857,
+      "step": 9775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.153709462461129e-06,
+      "loss": 0.0661,
+      "step": 9780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.14815637494447e-06,
+      "loss": 0.1088,
+      "step": 9785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.142603287427811e-06,
+      "loss": 0.0513,
+      "step": 9790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.137050199911152e-06,
+      "loss": 0.0719,
+      "step": 9795
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.131497112394491e-06,
+      "loss": 0.07,
+      "step": 9800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.125944024877833e-06,
+      "loss": 0.1544,
+      "step": 9805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.120390937361174e-06,
+      "loss": 0.2289,
+      "step": 9810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.114837849844515e-06,
+      "loss": 0.0585,
+      "step": 9815
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.109284762327854e-06,
+      "loss": 0.0721,
+      "step": 9820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.103731674811195e-06,
+      "loss": 0.0831,
+      "step": 9825
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.098178587294536e-06,
+      "loss": 0.0853,
+      "step": 9830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.092625499777877e-06,
+      "loss": 0.0464,
+      "step": 9835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.087072412261218e-06,
+      "loss": 0.0464,
+      "step": 9840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.08151932474456e-06,
+      "loss": 0.0639,
+      "step": 9845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.075966237227899e-06,
+      "loss": 0.0548,
+      "step": 9850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.07041314971124e-06,
+      "loss": 0.196,
+      "step": 9855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.06486006219458e-06,
+      "loss": 0.2379,
+      "step": 9860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.059306974677922e-06,
+      "loss": 0.0846,
+      "step": 9865
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.053753887161263e-06,
+      "loss": 0.0647,
+      "step": 9870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.048200799644602e-06,
+      "loss": 0.098,
+      "step": 9875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.042647712127945e-06,
+      "loss": 0.0939,
+      "step": 9880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.037094624611284e-06,
+      "loss": 0.0553,
+      "step": 9885
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.031541537094626e-06,
+      "loss": 0.0505,
+      "step": 9890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.025988449577967e-06,
+      "loss": 0.0596,
+      "step": 9895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.020435362061308e-06,
+      "loss": 0.097,
+      "step": 9900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.014882274544647e-06,
+      "loss": 0.1881,
+      "step": 9905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.009329187027988e-06,
+      "loss": 0.2562,
+      "step": 9910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.003776099511329e-06,
+      "loss": 0.0614,
+      "step": 9915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.99822301199467e-06,
+      "loss": 0.0803,
+      "step": 9920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.99266992447801e-06,
+      "loss": 0.0694,
+      "step": 9925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.987116836961352e-06,
+      "loss": 0.0807,
+      "step": 9930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.981563749444692e-06,
+      "loss": 0.0658,
+      "step": 9935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.976010661928033e-06,
+      "loss": 0.0545,
+      "step": 9940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.970457574411374e-06,
+      "loss": 0.0712,
+      "step": 9945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.964904486894715e-06,
+      "loss": 0.0791,
+      "step": 9950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.959351399378054e-06,
+      "loss": 0.1138,
+      "step": 9955
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.953798311861395e-06,
+      "loss": 0.2095,
+      "step": 9960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.948245224344736e-06,
+      "loss": 0.0778,
+      "step": 9965
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.942692136828077e-06,
+      "loss": 0.0691,
+      "step": 9970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.937139049311417e-06,
+      "loss": 0.0888,
+      "step": 9975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.931585961794758e-06,
+      "loss": 0.1236,
+      "step": 9980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.926032874278099e-06,
+      "loss": 0.0293,
+      "step": 9985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.92047978676144e-06,
+      "loss": 0.0301,
+      "step": 9990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.914926699244781e-06,
+      "loss": 0.057,
+      "step": 9995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.909373611728122e-06,
+      "loss": 0.083,
+      "step": 10000
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.033624935895204544,
+      "eval_runtime": 1824.5042,
+      "eval_samples_per_second": 8.779,
+      "eval_steps_per_second": 2.195,
+      "eval_wer": 0.08218535881716553,
+      "step": 10000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.903820524211463e-06,
+      "loss": 0.1404,
+      "step": 10005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.898267436694803e-06,
+      "loss": 0.2263,
+      "step": 10010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.892714349178144e-06,
+      "loss": 0.0728,
+      "step": 10015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.887161261661485e-06,
+      "loss": 0.0705,
+      "step": 10020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.881608174144826e-06,
+      "loss": 0.0972,
+      "step": 10025
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.876055086628165e-06,
+      "loss": 0.0948,
+      "step": 10030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.870501999111508e-06,
+      "loss": 0.0245,
+      "step": 10035
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.864948911594847e-06,
+      "loss": 0.0558,
+      "step": 10040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.859395824078188e-06,
+      "loss": 0.0816,
+      "step": 10045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.85384273656153e-06,
+      "loss": 0.1294,
+      "step": 10050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.84828964904487e-06,
+      "loss": 0.2248,
+      "step": 10055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.84273656152821e-06,
+      "loss": 0.2146,
+      "step": 10060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.83718347401155e-06,
+      "loss": 0.0828,
+      "step": 10065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.831630386494892e-06,
+      "loss": 0.0751,
+      "step": 10070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.826077298978233e-06,
+      "loss": 0.0982,
+      "step": 10075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.820524211461572e-06,
+      "loss": 0.1165,
+      "step": 10080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.814971123944915e-06,
+      "loss": 0.0855,
+      "step": 10085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.809418036428254e-06,
+      "loss": 0.0588,
+      "step": 10090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.803864948911596e-06,
+      "loss": 0.0708,
+      "step": 10095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.798311861394937e-06,
+      "loss": 0.1188,
+      "step": 10100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.792758773878278e-06,
+      "loss": 0.1458,
+      "step": 10105
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.787205686361617e-06,
+      "loss": 0.3549,
+      "step": 10110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.781652598844958e-06,
+      "loss": 0.051,
+      "step": 10115
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.776099511328299e-06,
+      "loss": 0.0744,
+      "step": 10120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.77054642381164e-06,
+      "loss": 0.0744,
+      "step": 10125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.764993336294981e-06,
+      "loss": 0.0964,
+      "step": 10130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.75944024877832e-06,
+      "loss": 0.0551,
+      "step": 10135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.753887161261663e-06,
+      "loss": 0.041,
+      "step": 10140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.748334073745003e-06,
+      "loss": 0.0542,
+      "step": 10145
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.742780986228344e-06,
+      "loss": 0.1429,
+      "step": 10150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.737227898711685e-06,
+      "loss": 0.1656,
+      "step": 10155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.731674811195026e-06,
+      "loss": 0.242,
+      "step": 10160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.726121723678365e-06,
+      "loss": 0.0652,
+      "step": 10165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.720568636161706e-06,
+      "loss": 0.0791,
+      "step": 10170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.715015548645047e-06,
+      "loss": 0.0712,
+      "step": 10175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.709462461128388e-06,
+      "loss": 0.1123,
+      "step": 10180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.703909373611728e-06,
+      "loss": 0.0367,
+      "step": 10185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.69835628609507e-06,
+      "loss": 0.0754,
+      "step": 10190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.69280319857841e-06,
+      "loss": 0.0435,
+      "step": 10195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.687250111061751e-06,
+      "loss": 0.1014,
+      "step": 10200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.681697023545092e-06,
+      "loss": 0.1813,
+      "step": 10205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.676143936028433e-06,
+      "loss": 0.2386,
+      "step": 10210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.670590848511773e-06,
+      "loss": 0.0901,
+      "step": 10215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.665037760995114e-06,
+      "loss": 0.0924,
+      "step": 10220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.659484673478455e-06,
+      "loss": 0.0571,
+      "step": 10225
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.653931585961796e-06,
+      "loss": 0.0835,
+      "step": 10230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.648378498445135e-06,
+      "loss": 0.0545,
+      "step": 10235
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.642825410928478e-06,
+      "loss": 0.022,
+      "step": 10240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.637272323411817e-06,
+      "loss": 0.0543,
+      "step": 10245
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.631719235895158e-06,
+      "loss": 0.0893,
+      "step": 10250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.6261661483785e-06,
+      "loss": 0.1344,
+      "step": 10255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.62061306086184e-06,
+      "loss": 0.2911,
+      "step": 10260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.61505997334518e-06,
+      "loss": 0.0561,
+      "step": 10265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.60950688582852e-06,
+      "loss": 0.0803,
+      "step": 10270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.603953798311862e-06,
+      "loss": 0.0646,
+      "step": 10275
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.598400710795203e-06,
+      "loss": 0.0943,
+      "step": 10280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.592847623278544e-06,
+      "loss": 0.055,
+      "step": 10285
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.587294535761883e-06,
+      "loss": 0.0501,
+      "step": 10290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.581741448245226e-06,
+      "loss": 0.0944,
+      "step": 10295
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.576188360728566e-06,
+      "loss": 0.0865,
+      "step": 10300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.570635273211907e-06,
+      "loss": 0.1229,
+      "step": 10305
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.565082185695248e-06,
+      "loss": 0.2852,
+      "step": 10310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.559529098178589e-06,
+      "loss": 0.0619,
+      "step": 10315
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.553976010661928e-06,
+      "loss": 0.074,
+      "step": 10320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.548422923145269e-06,
+      "loss": 0.0861,
+      "step": 10325
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.54286983562861e-06,
+      "loss": 0.0899,
+      "step": 10330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.537316748111951e-06,
+      "loss": 0.0766,
+      "step": 10335
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.53176366059529e-06,
+      "loss": 0.0935,
+      "step": 10340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.526210573078633e-06,
+      "loss": 0.0629,
+      "step": 10345
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.520657485561973e-06,
+      "loss": 0.0753,
+      "step": 10350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.515104398045314e-06,
+      "loss": 0.1312,
+      "step": 10355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.509551310528655e-06,
+      "loss": 0.324,
+      "step": 10360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.503998223011996e-06,
+      "loss": 0.069,
+      "step": 10365
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.498445135495335e-06,
+      "loss": 0.0876,
+      "step": 10370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.492892047978676e-06,
+      "loss": 0.0953,
+      "step": 10375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.487338960462017e-06,
+      "loss": 0.0984,
+      "step": 10380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.481785872945358e-06,
+      "loss": 0.0535,
+      "step": 10385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.476232785428698e-06,
+      "loss": 0.0803,
+      "step": 10390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.47067969791204e-06,
+      "loss": 0.0545,
+      "step": 10395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.46512661039538e-06,
+      "loss": 0.104,
+      "step": 10400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.459573522878721e-06,
+      "loss": 0.1661,
+      "step": 10405
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.454020435362062e-06,
+      "loss": 0.2866,
+      "step": 10410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.448467347845403e-06,
+      "loss": 0.0697,
+      "step": 10415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.442914260328744e-06,
+      "loss": 0.0643,
+      "step": 10420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.437361172812084e-06,
+      "loss": 0.0958,
+      "step": 10425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.431808085295425e-06,
+      "loss": 0.0842,
+      "step": 10430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.426254997778766e-06,
+      "loss": 0.0644,
+      "step": 10435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.420701910262107e-06,
+      "loss": 0.0514,
+      "step": 10440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.415148822745446e-06,
+      "loss": 0.0719,
+      "step": 10445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.409595735228789e-06,
+      "loss": 0.1048,
+      "step": 10450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.404042647712128e-06,
+      "loss": 0.1167,
+      "step": 10455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.39848956019547e-06,
+      "loss": 0.2688,
+      "step": 10460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.39293647267881e-06,
+      "loss": 0.0667,
+      "step": 10465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.387383385162151e-06,
+      "loss": 0.0867,
+      "step": 10470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.38183029764549e-06,
+      "loss": 0.0938,
+      "step": 10475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.376277210128832e-06,
+      "loss": 0.0971,
+      "step": 10480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.370724122612173e-06,
+      "loss": 0.0371,
+      "step": 10485
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.365171035095514e-06,
+      "loss": 0.0456,
+      "step": 10490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.359617947578853e-06,
+      "loss": 0.0575,
+      "step": 10495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.354064860062196e-06,
+      "loss": 0.1993,
+      "step": 10500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.348511772545536e-06,
+      "loss": 0.1699,
+      "step": 10505
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.342958685028877e-06,
+      "loss": 0.267,
+      "step": 10510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.337405597512218e-06,
+      "loss": 0.0599,
+      "step": 10515
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.331852509995559e-06,
+      "loss": 0.0663,
+      "step": 10520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.326299422478898e-06,
+      "loss": 0.0596,
+      "step": 10525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.320746334962239e-06,
+      "loss": 0.0873,
+      "step": 10530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.31519324744558e-06,
+      "loss": 0.0259,
+      "step": 10535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.309640159928921e-06,
+      "loss": 0.0751,
+      "step": 10540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.304087072412262e-06,
+      "loss": 0.057,
+      "step": 10545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.298533984895603e-06,
+      "loss": 0.093,
+      "step": 10550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.292980897378944e-06,
+      "loss": 0.0977,
+      "step": 10555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.287427809862284e-06,
+      "loss": 0.239,
+      "step": 10560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.281874722345625e-06,
+      "loss": 0.0785,
+      "step": 10565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.276321634828966e-06,
+      "loss": 0.0693,
+      "step": 10570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.270768547312307e-06,
+      "loss": 0.0745,
+      "step": 10575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.265215459795646e-06,
+      "loss": 0.1049,
+      "step": 10580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.259662372278987e-06,
+      "loss": 0.0331,
+      "step": 10585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.254109284762328e-06,
+      "loss": 0.0817,
+      "step": 10590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.24855619724567e-06,
+      "loss": 0.0558,
+      "step": 10595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.243003109729009e-06,
+      "loss": 0.0921,
+      "step": 10600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.237450022212352e-06,
+      "loss": 0.228,
+      "step": 10605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.231896934695691e-06,
+      "loss": 0.1868,
+      "step": 10610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.226343847179032e-06,
+      "loss": 0.0731,
+      "step": 10615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.220790759662373e-06,
+      "loss": 0.0744,
+      "step": 10620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.215237672145714e-06,
+      "loss": 0.0719,
+      "step": 10625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.209684584629054e-06,
+      "loss": 0.0821,
+      "step": 10630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.204131497112395e-06,
+      "loss": 0.0547,
+      "step": 10635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.198578409595736e-06,
+      "loss": 0.0501,
+      "step": 10640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.193025322079077e-06,
+      "loss": 0.054,
+      "step": 10645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.187472234562416e-06,
+      "loss": 0.0934,
+      "step": 10650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.181919147045759e-06,
+      "loss": 0.189,
+      "step": 10655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.176366059529098e-06,
+      "loss": 0.1917,
+      "step": 10660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.17081297201244e-06,
+      "loss": 0.0897,
+      "step": 10665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.16525988449578e-06,
+      "loss": 0.0714,
+      "step": 10670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.159706796979121e-06,
+      "loss": 0.0775,
+      "step": 10675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.154153709462463e-06,
+      "loss": 0.098,
+      "step": 10680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.148600621945802e-06,
+      "loss": 0.0518,
+      "step": 10685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.143047534429143e-06,
+      "loss": 0.0568,
+      "step": 10690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.137494446912484e-06,
+      "loss": 0.069,
+      "step": 10695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.131941359395825e-06,
+      "loss": 0.0887,
+      "step": 10700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.126388271879166e-06,
+      "loss": 0.124,
+      "step": 10705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.120835184362507e-06,
+      "loss": 0.249,
+      "step": 10710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.115282096845847e-06,
+      "loss": 0.0698,
+      "step": 10715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.109729009329188e-06,
+      "loss": 0.0587,
+      "step": 10720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.104175921812529e-06,
+      "loss": 0.0617,
+      "step": 10725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.09862283429587e-06,
+      "loss": 0.1027,
+      "step": 10730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.093069746779209e-06,
+      "loss": 0.0704,
+      "step": 10735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.08751665926255e-06,
+      "loss": 0.0428,
+      "step": 10740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.081963571745891e-06,
+      "loss": 0.0484,
+      "step": 10745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.076410484229232e-06,
+      "loss": 0.1133,
+      "step": 10750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.070857396712572e-06,
+      "loss": 0.1798,
+      "step": 10755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.065304309195914e-06,
+      "loss": 0.2342,
+      "step": 10760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.059751221679254e-06,
+      "loss": 0.1101,
+      "step": 10765
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.054198134162595e-06,
+      "loss": 0.0627,
+      "step": 10770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.048645046645936e-06,
+      "loss": 0.0895,
+      "step": 10775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.043091959129277e-06,
+      "loss": 0.0515,
+      "step": 10780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.037538871612616e-06,
+      "loss": 0.0282,
+      "step": 10785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.031985784095957e-06,
+      "loss": 0.0678,
+      "step": 10790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.026432696579298e-06,
+      "loss": 0.0769,
+      "step": 10795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.02087960906264e-06,
+      "loss": 0.078,
+      "step": 10800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.015326521545979e-06,
+      "loss": 0.2172,
+      "step": 10805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.009773434029322e-06,
+      "loss": 0.3175,
+      "step": 10810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.004220346512663e-06,
+      "loss": 0.064,
+      "step": 10815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.998667258996002e-06,
+      "loss": 0.0797,
+      "step": 10820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.993114171479343e-06,
+      "loss": 0.061,
+      "step": 10825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.987561083962684e-06,
+      "loss": 0.0779,
+      "step": 10830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.982007996446025e-06,
+      "loss": 0.0508,
+      "step": 10835
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.976454908929365e-06,
+      "loss": 0.0208,
+      "step": 10840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.970901821412706e-06,
+      "loss": 0.069,
+      "step": 10845
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.965348733896047e-06,
+      "loss": 0.1362,
+      "step": 10850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.959795646379388e-06,
+      "loss": 0.1576,
+      "step": 10855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.954242558862729e-06,
+      "loss": 0.245,
+      "step": 10860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.94868947134607e-06,
+      "loss": 0.0603,
+      "step": 10865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.94313638382941e-06,
+      "loss": 0.0738,
+      "step": 10870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.93758329631275e-06,
+      "loss": 0.0873,
+      "step": 10875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.932030208796091e-06,
+      "loss": 0.1042,
+      "step": 10880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.926477121279433e-06,
+      "loss": 0.0486,
+      "step": 10885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.920924033762772e-06,
+      "loss": 0.0378,
+      "step": 10890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.915370946246113e-06,
+      "loss": 0.0678,
+      "step": 10895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.909817858729454e-06,
+      "loss": 0.087,
+      "step": 10900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.904264771212795e-06,
+      "loss": 0.1775,
+      "step": 10905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.898711683696134e-06,
+      "loss": 0.2822,
+      "step": 10910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.893158596179477e-06,
+      "loss": 0.0672,
+      "step": 10915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.887605508662817e-06,
+      "loss": 0.0687,
+      "step": 10920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.882052421146158e-06,
+      "loss": 0.0712,
+      "step": 10925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.876499333629499e-06,
+      "loss": 0.0584,
+      "step": 10930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.87094624611284e-06,
+      "loss": 0.0539,
+      "step": 10935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.865393158596179e-06,
+      "loss": 0.0178,
+      "step": 10940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.85984007107952e-06,
+      "loss": 0.0817,
+      "step": 10945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.854286983562863e-06,
+      "loss": 0.1031,
+      "step": 10950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.848733896046202e-06,
+      "loss": 0.1493,
+      "step": 10955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.843180808529543e-06,
+      "loss": 0.3111,
+      "step": 10960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.837627721012884e-06,
+      "loss": 0.0735,
+      "step": 10965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.832074633496226e-06,
+      "loss": 0.0746,
+      "step": 10970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.826521545979565e-06,
+      "loss": 0.0674,
+      "step": 10975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.820968458462906e-06,
+      "loss": 0.0793,
+      "step": 10980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.815415370946247e-06,
+      "loss": 0.0661,
+      "step": 10985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.809862283429588e-06,
+      "loss": 0.0174,
+      "step": 10990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.804309195912927e-06,
+      "loss": 0.0432,
+      "step": 10995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.798756108396268e-06,
+      "loss": 0.0964,
+      "step": 11000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.79320302087961e-06,
+      "loss": 0.1417,
+      "step": 11005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.78764993336295e-06,
+      "loss": 0.2896,
+      "step": 11010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.782096845846292e-06,
+      "loss": 0.0701,
+      "step": 11015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.776543758329633e-06,
+      "loss": 0.0699,
+      "step": 11020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.770990670812972e-06,
+      "loss": 0.0587,
+      "step": 11025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.765437583296313e-06,
+      "loss": 0.0731,
+      "step": 11030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.759884495779654e-06,
+      "loss": 0.054,
+      "step": 11035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.754331408262995e-06,
+      "loss": 0.0428,
+      "step": 11040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.748778320746335e-06,
+      "loss": 0.072,
+      "step": 11045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.743225233229676e-06,
+      "loss": 0.1259,
+      "step": 11050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.737672145713017e-06,
+      "loss": 0.077,
+      "step": 11055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.732119058196358e-06,
+      "loss": 0.1818,
+      "step": 11060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.726565970679697e-06,
+      "loss": 0.0632,
+      "step": 11065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.72101288316304e-06,
+      "loss": 0.0606,
+      "step": 11070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.71545979564638e-06,
+      "loss": 0.0774,
+      "step": 11075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.70990670812972e-06,
+      "loss": 0.0928,
+      "step": 11080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.704353620613061e-06,
+      "loss": 0.0493,
+      "step": 11085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.698800533096403e-06,
+      "loss": 0.074,
+      "step": 11090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.693247445579744e-06,
+      "loss": 0.0717,
+      "step": 11095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.687694358063083e-06,
+      "loss": 0.0711,
+      "step": 11100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.682141270546426e-06,
+      "loss": 0.1324,
+      "step": 11105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.676588183029765e-06,
+      "loss": 0.2292,
+      "step": 11110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.671035095513106e-06,
+      "loss": 0.0606,
+      "step": 11115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.665482007996447e-06,
+      "loss": 0.0759,
+      "step": 11120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.659928920479788e-06,
+      "loss": 0.0844,
+      "step": 11125
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.654375832963128e-06,
+      "loss": 0.0892,
+      "step": 11130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.648822745446469e-06,
+      "loss": 0.066,
+      "step": 11135
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.64326965792981e-06,
+      "loss": 0.0444,
+      "step": 11140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.63771657041315e-06,
+      "loss": 0.0964,
+      "step": 11145
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.63216348289649e-06,
+      "loss": 0.087,
+      "step": 11150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.626610395379832e-06,
+      "loss": 0.2088,
+      "step": 11155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.621057307863172e-06,
+      "loss": 0.3376,
+      "step": 11160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.615504220346513e-06,
+      "loss": 0.058,
+      "step": 11165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.609951132829854e-06,
+      "loss": 0.0705,
+      "step": 11170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.604398045313195e-06,
+      "loss": 0.0804,
+      "step": 11175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.598844957796535e-06,
+      "loss": 0.0868,
+      "step": 11180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.593291870279876e-06,
+      "loss": 0.0629,
+      "step": 11185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.587738782763216e-06,
+      "loss": 0.0387,
+      "step": 11190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.582185695246558e-06,
+      "loss": 0.0682,
+      "step": 11195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.576632607729898e-06,
+      "loss": 0.0925,
+      "step": 11200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.571079520213239e-06,
+      "loss": 0.1252,
+      "step": 11205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.5655264326965796e-06,
+      "loss": 0.2454,
+      "step": 11210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.559973345179921e-06,
+      "loss": 0.0921,
+      "step": 11215
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.554420257663262e-06,
+      "loss": 0.0481,
+      "step": 11220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.548867170146602e-06,
+      "loss": 0.0748,
+      "step": 11225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.543314082629943e-06,
+      "loss": 0.0775,
+      "step": 11230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.537760995113283e-06,
+      "loss": 0.0666,
+      "step": 11235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.532207907596625e-06,
+      "loss": 0.0388,
+      "step": 11240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.526654820079965e-06,
+      "loss": 0.0551,
+      "step": 11245
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.521101732563306e-06,
+      "loss": 0.087,
+      "step": 11250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5155486450466466e-06,
+      "loss": 0.1213,
+      "step": 11255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.509995557529988e-06,
+      "loss": 0.2799,
+      "step": 11260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.504442470013328e-06,
+      "loss": 0.0806,
+      "step": 11265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.498889382496669e-06,
+      "loss": 0.0787,
+      "step": 11270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.493336294980009e-06,
+      "loss": 0.0837,
+      "step": 11275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.48778320746335e-06,
+      "loss": 0.0846,
+      "step": 11280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.48223011994669e-06,
+      "loss": 0.0462,
+      "step": 11285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.476677032430032e-06,
+      "loss": 0.0767,
+      "step": 11290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.4711239449133725e-06,
+      "loss": 0.0777,
+      "step": 11295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.465570857396714e-06,
+      "loss": 0.099,
+      "step": 11300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.460017769880054e-06,
+      "loss": 0.163,
+      "step": 11305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.454464682363395e-06,
+      "loss": 0.2873,
+      "step": 11310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.448911594846735e-06,
+      "loss": 0.0684,
+      "step": 11315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.443358507330076e-06,
+      "loss": 0.0565,
+      "step": 11320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.437805419813416e-06,
+      "loss": 0.0902,
+      "step": 11325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.432252332296757e-06,
+      "loss": 0.0835,
+      "step": 11330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.426699244780098e-06,
+      "loss": 0.0621,
+      "step": 11335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.421146157263439e-06,
+      "loss": 0.0421,
+      "step": 11340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.415593069746779e-06,
+      "loss": 0.0957,
+      "step": 11345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.410039982230121e-06,
+      "loss": 0.0845,
+      "step": 11350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.404486894713462e-06,
+      "loss": 0.1433,
+      "step": 11355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.398933807196802e-06,
+      "loss": 0.1906,
+      "step": 11360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.393380719680143e-06,
+      "loss": 0.0645,
+      "step": 11365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.387827632163483e-06,
+      "loss": 0.0532,
+      "step": 11370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.3822745446468244e-06,
+      "loss": 0.0664,
+      "step": 11375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.376721457130165e-06,
+      "loss": 0.0747,
+      "step": 11380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.371168369613506e-06,
+      "loss": 0.0381,
+      "step": 11385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.365615282096846e-06,
+      "loss": 0.0541,
+      "step": 11390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.360062194580188e-06,
+      "loss": 0.0625,
+      "step": 11395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.354509107063528e-06,
+      "loss": 0.0858,
+      "step": 11400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.348956019546869e-06,
+      "loss": 0.1765,
+      "step": 11405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.343402932030209e-06,
+      "loss": 0.3694,
+      "step": 11410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.33784984451355e-06,
+      "loss": 0.0708,
+      "step": 11415
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.332296756996891e-06,
+      "loss": 0.0632,
+      "step": 11420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.326743669480232e-06,
+      "loss": 0.0825,
+      "step": 11425
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.321190581963572e-06,
+      "loss": 0.0794,
+      "step": 11430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.315637494446913e-06,
+      "loss": 0.0507,
+      "step": 11435
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.310084406930253e-06,
+      "loss": 0.072,
+      "step": 11440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.304531319413595e-06,
+      "loss": 0.049,
+      "step": 11445
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.298978231896935e-06,
+      "loss": 0.0754,
+      "step": 11450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.293425144380276e-06,
+      "loss": 0.139,
+      "step": 11455
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.2878720568636166e-06,
+      "loss": 0.21,
+      "step": 11460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.282318969346958e-06,
+      "loss": 0.0622,
+      "step": 11465
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.276765881830298e-06,
+      "loss": 0.0679,
+      "step": 11470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.271212794313639e-06,
+      "loss": 0.0903,
+      "step": 11475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.265659706796979e-06,
+      "loss": 0.0885,
+      "step": 11480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.26010661928032e-06,
+      "loss": 0.0284,
+      "step": 11485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.254553531763662e-06,
+      "loss": 0.0431,
+      "step": 11490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.2490004442470015e-06,
+      "loss": 0.0551,
+      "step": 11495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.243447356730343e-06,
+      "loss": 0.1224,
+      "step": 11500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.237894269213684e-06,
+      "loss": 0.1565,
+      "step": 11505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.232341181697025e-06,
+      "loss": 0.24,
+      "step": 11510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.226788094180365e-06,
+      "loss": 0.077,
+      "step": 11515
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.221235006663706e-06,
+      "loss": 0.0688,
+      "step": 11520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.215681919147046e-06,
+      "loss": 0.0758,
+      "step": 11525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.210128831630387e-06,
+      "loss": 0.1004,
+      "step": 11530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.2045757441137274e-06,
+      "loss": 0.0554,
+      "step": 11535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.1990226565970685e-06,
+      "loss": 0.0266,
+      "step": 11540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.193469569080409e-06,
+      "loss": 0.0425,
+      "step": 11545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.187916481563751e-06,
+      "loss": 0.0686,
+      "step": 11550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.182363394047091e-06,
+      "loss": 0.1635,
+      "step": 11555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.176810306530432e-06,
+      "loss": 0.2836,
+      "step": 11560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.171257219013772e-06,
+      "loss": 0.0675,
+      "step": 11565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.165704131497113e-06,
+      "loss": 0.0556,
+      "step": 11570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.160151043980453e-06,
+      "loss": 0.0896,
+      "step": 11575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1545979564637944e-06,
+      "loss": 0.1134,
+      "step": 11580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.149044868947135e-06,
+      "loss": 0.0585,
+      "step": 11585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.143491781430476e-06,
+      "loss": 0.0608,
+      "step": 11590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.137938693913816e-06,
+      "loss": 0.0798,
+      "step": 11595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.132385606397158e-06,
+      "loss": 0.0943,
+      "step": 11600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.126832518880498e-06,
+      "loss": 0.1184,
+      "step": 11605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.121279431363839e-06,
+      "loss": 0.1871,
+      "step": 11610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.115726343847179e-06,
+      "loss": 0.0624,
+      "step": 11615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.11017325633052e-06,
+      "loss": 0.0699,
+      "step": 11620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1046201688138615e-06,
+      "loss": 0.0656,
+      "step": 11625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.099067081297202e-06,
+      "loss": 0.0663,
+      "step": 11630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.093513993780543e-06,
+      "loss": 0.0486,
+      "step": 11635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.087960906263883e-06,
+      "loss": 0.052,
+      "step": 11640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.082407818747225e-06,
+      "loss": 0.1055,
+      "step": 11645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.076854731230564e-06,
+      "loss": 0.0685,
+      "step": 11650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.071301643713906e-06,
+      "loss": 0.114,
+      "step": 11655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.065748556197246e-06,
+      "loss": 0.2795,
+      "step": 11660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.060195468680587e-06,
+      "loss": 0.1081,
+      "step": 11665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.054642381163928e-06,
+      "loss": 0.056,
+      "step": 11670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.049089293647269e-06,
+      "loss": 0.0806,
+      "step": 11675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.043536206130609e-06,
+      "loss": 0.0957,
+      "step": 11680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.03798311861395e-06,
+      "loss": 0.0456,
+      "step": 11685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.03243003109729e-06,
+      "loss": 0.0738,
+      "step": 11690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.026876943580631e-06,
+      "loss": 0.082,
+      "step": 11695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0213238560639715e-06,
+      "loss": 0.0751,
+      "step": 11700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.015770768547313e-06,
+      "loss": 0.1192,
+      "step": 11705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.010217681030654e-06,
+      "loss": 0.2127,
+      "step": 11710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.004664593513995e-06,
+      "loss": 0.042,
+      "step": 11715
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.999111505997335e-06,
+      "loss": 0.0678,
+      "step": 11720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.993558418480676e-06,
+      "loss": 0.0663,
+      "step": 11725
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.988005330964016e-06,
+      "loss": 0.0781,
+      "step": 11730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.982452243447357e-06,
+      "loss": 0.077,
+      "step": 11735
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.9768991559306974e-06,
+      "loss": 0.0657,
+      "step": 11740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.9713460684140385e-06,
+      "loss": 0.0659,
+      "step": 11745
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.965792980897379e-06,
+      "loss": 0.0878,
+      "step": 11750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.960239893380721e-06,
+      "loss": 0.0875,
+      "step": 11755
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.954686805864062e-06,
+      "loss": 0.1752,
+      "step": 11760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.949133718347402e-06,
+      "loss": 0.0628,
+      "step": 11765
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.943580630830743e-06,
+      "loss": 0.0576,
+      "step": 11770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.938027543314083e-06,
+      "loss": 0.0735,
+      "step": 11775
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.932474455797424e-06,
+      "loss": 0.0748,
+      "step": 11780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9269213682807644e-06,
+      "loss": 0.046,
+      "step": 11785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9213682807641055e-06,
+      "loss": 0.0389,
+      "step": 11790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.915815193247446e-06,
+      "loss": 0.0605,
+      "step": 11795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.910262105730788e-06,
+      "loss": 0.0961,
+      "step": 11800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.904709018214127e-06,
+      "loss": 0.1166,
+      "step": 11805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.899155930697469e-06,
+      "loss": 0.258,
+      "step": 11810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.893602843180809e-06,
+      "loss": 0.0542,
+      "step": 11815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.88804975566415e-06,
+      "loss": 0.0674,
+      "step": 11820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.88249666814749e-06,
+      "loss": 0.0864,
+      "step": 11825
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.8769435806308315e-06,
+      "loss": 0.0839,
+      "step": 11830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.871390493114172e-06,
+      "loss": 0.0785,
+      "step": 11835
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.865837405597513e-06,
+      "loss": 0.0333,
+      "step": 11840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.860284318080853e-06,
+      "loss": 0.0789,
+      "step": 11845
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.854731230564194e-06,
+      "loss": 0.0888,
+      "step": 11850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.849178143047534e-06,
+      "loss": 0.1334,
+      "step": 11855
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.843625055530876e-06,
+      "loss": 0.2851,
+      "step": 11860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.838071968014216e-06,
+      "loss": 0.0718,
+      "step": 11865
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.832518880497557e-06,
+      "loss": 0.0622,
+      "step": 11870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.826965792980898e-06,
+      "loss": 0.0734,
+      "step": 11875
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.821412705464239e-06,
+      "loss": 0.0712,
+      "step": 11880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.815859617947579e-06,
+      "loss": 0.0618,
+      "step": 11885
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.81030653043092e-06,
+      "loss": 0.0463,
+      "step": 11890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.804753442914261e-06,
+      "loss": 0.0656,
+      "step": 11895
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.799200355397601e-06,
+      "loss": 0.1304,
+      "step": 11900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.793647267880943e-06,
+      "loss": 0.1738,
+      "step": 11905
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.788094180364283e-06,
+      "loss": 0.3324,
+      "step": 11910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.7825410928476244e-06,
+      "loss": 0.0565,
+      "step": 11915
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.776988005330965e-06,
+      "loss": 0.0709,
+      "step": 11920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.771434917814306e-06,
+      "loss": 0.0738,
+      "step": 11925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.765881830297646e-06,
+      "loss": 0.0898,
+      "step": 11930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.760328742780987e-06,
+      "loss": 0.0875,
+      "step": 11935
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.754775655264327e-06,
+      "loss": 0.0233,
+      "step": 11940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.749222567747668e-06,
+      "loss": 0.0473,
+      "step": 11945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.7436694802310085e-06,
+      "loss": 0.1188,
+      "step": 11950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.73811639271435e-06,
+      "loss": 0.1197,
+      "step": 11955
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.73256330519769e-06,
+      "loss": 0.2622,
+      "step": 11960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.727010217681032e-06,
+      "loss": 0.0541,
+      "step": 11965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.721457130164372e-06,
+      "loss": 0.0638,
+      "step": 11970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.715904042647713e-06,
+      "loss": 0.059,
+      "step": 11975
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.710350955131053e-06,
+      "loss": 0.1019,
+      "step": 11980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.704797867614394e-06,
+      "loss": 0.0312,
+      "step": 11985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.6992447800977344e-06,
+      "loss": 0.0307,
+      "step": 11990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.6936916925810755e-06,
+      "loss": 0.0525,
+      "step": 11995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.688138605064416e-06,
+      "loss": 0.1069,
+      "step": 12000
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.03537467494606972,
+      "eval_runtime": 1821.7996,
+      "eval_samples_per_second": 8.792,
+      "eval_steps_per_second": 2.198,
+      "eval_wer": 0.07814641182834475,
+      "step": 12000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.682585517547757e-06,
+      "loss": 0.1494,
+      "step": 12005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.677032430031097e-06,
+      "loss": 0.287,
+      "step": 12010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.671479342514439e-06,
+      "loss": 0.051,
+      "step": 12015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.665926254997779e-06,
+      "loss": 0.0788,
+      "step": 12020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.66037316748112e-06,
+      "loss": 0.0693,
+      "step": 12025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.654820079964461e-06,
+      "loss": 0.0857,
+      "step": 12030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.6492669924478015e-06,
+      "loss": 0.0516,
+      "step": 12035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.6437139049311425e-06,
+      "loss": 0.1087,
+      "step": 12040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.638160817414483e-06,
+      "loss": 0.0696,
+      "step": 12045
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.632607729897824e-06,
+      "loss": 0.1212,
+      "step": 12050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.627054642381164e-06,
+      "loss": 0.1714,
+      "step": 12055
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.621501554864506e-06,
+      "loss": 0.2119,
+      "step": 12060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.615948467347846e-06,
+      "loss": 0.0623,
+      "step": 12065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.610395379831187e-06,
+      "loss": 0.0845,
+      "step": 12070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.6048422923145274e-06,
+      "loss": 0.0777,
+      "step": 12075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.5992892047978685e-06,
+      "loss": 0.0991,
+      "step": 12080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.593736117281209e-06,
+      "loss": 0.0411,
+      "step": 12085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.58818302976455e-06,
+      "loss": 0.0513,
+      "step": 12090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.58262994224789e-06,
+      "loss": 0.0582,
+      "step": 12095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.577076854731231e-06,
+      "loss": 0.0775,
+      "step": 12100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.571523767214571e-06,
+      "loss": 0.1754,
+      "step": 12105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.565970679697913e-06,
+      "loss": 0.2479,
+      "step": 12110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.5604175921812525e-06,
+      "loss": 0.0757,
+      "step": 12115
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5548645046645944e-06,
+      "loss": 0.0696,
+      "step": 12120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.549311417147935e-06,
+      "loss": 0.0771,
+      "step": 12125
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.543758329631276e-06,
+      "loss": 0.0973,
+      "step": 12130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.538205242114616e-06,
+      "loss": 0.0523,
+      "step": 12135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.532652154597957e-06,
+      "loss": 0.0466,
+      "step": 12140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.527099067081297e-06,
+      "loss": 0.061,
+      "step": 12145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.521545979564638e-06,
+      "loss": 0.1244,
+      "step": 12150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5159928920479785e-06,
+      "loss": 0.1754,
+      "step": 12155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5104398045313196e-06,
+      "loss": 0.2873,
+      "step": 12160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5048867170146615e-06,
+      "loss": 0.0712,
+      "step": 12165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.499333629498002e-06,
+      "loss": 0.0774,
+      "step": 12170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.493780541981343e-06,
+      "loss": 0.074,
+      "step": 12175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.488227454464683e-06,
+      "loss": 0.0728,
+      "step": 12180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.482674366948024e-06,
+      "loss": 0.0781,
+      "step": 12185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.477121279431364e-06,
+      "loss": 0.0659,
+      "step": 12190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.471568191914705e-06,
+      "loss": 0.0685,
+      "step": 12195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.4660151043980455e-06,
+      "loss": 0.1184,
+      "step": 12200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.4604620168813866e-06,
+      "loss": 0.1693,
+      "step": 12205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.454908929364727e-06,
+      "loss": 0.2137,
+      "step": 12210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.449355841848069e-06,
+      "loss": 0.0506,
+      "step": 12215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.443802754331409e-06,
+      "loss": 0.0757,
+      "step": 12220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.43824966681475e-06,
+      "loss": 0.0496,
+      "step": 12225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.43269657929809e-06,
+      "loss": 0.0779,
+      "step": 12230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.427143491781431e-06,
+      "loss": 0.0398,
+      "step": 12235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4215904042647715e-06,
+      "loss": 0.0343,
+      "step": 12240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4160373167481125e-06,
+      "loss": 0.0458,
+      "step": 12245
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.410484229231453e-06,
+      "loss": 0.1117,
+      "step": 12250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.404931141714794e-06,
+      "loss": 0.164,
+      "step": 12255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.399378054198134e-06,
+      "loss": 0.2337,
+      "step": 12260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.393824966681476e-06,
+      "loss": 0.0651,
+      "step": 12265
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.388271879164816e-06,
+      "loss": 0.0593,
+      "step": 12270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.382718791648157e-06,
+      "loss": 0.0968,
+      "step": 12275
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.3771657041314974e-06,
+      "loss": 0.0815,
+      "step": 12280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.3716126166148385e-06,
+      "loss": 0.0484,
+      "step": 12285
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.366059529098179e-06,
+      "loss": 0.0361,
+      "step": 12290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.36050644158152e-06,
+      "loss": 0.0624,
+      "step": 12295
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.354953354064861e-06,
+      "loss": 0.0901,
+      "step": 12300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.349400266548201e-06,
+      "loss": 0.1173,
+      "step": 12305
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.343847179031543e-06,
+      "loss": 0.2258,
+      "step": 12310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.338294091514882e-06,
+      "loss": 0.0756,
+      "step": 12315
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.332741003998224e-06,
+      "loss": 0.0601,
+      "step": 12320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3271879164815644e-06,
+      "loss": 0.0642,
+      "step": 12325
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3216348289649055e-06,
+      "loss": 0.1048,
+      "step": 12330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.316081741448246e-06,
+      "loss": 0.0456,
+      "step": 12335
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.310528653931587e-06,
+      "loss": 0.028,
+      "step": 12340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.304975566414927e-06,
+      "loss": 0.0515,
+      "step": 12345
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.299422478898268e-06,
+      "loss": 0.0595,
+      "step": 12350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.293869391381608e-06,
+      "loss": 0.1668,
+      "step": 12355
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.288316303864949e-06,
+      "loss": 0.1696,
+      "step": 12360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.2827632163482896e-06,
+      "loss": 0.0524,
+      "step": 12365
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.2772101288316315e-06,
+      "loss": 0.0612,
+      "step": 12370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.271657041314972e-06,
+      "loss": 0.0822,
+      "step": 12375
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.266103953798313e-06,
+      "loss": 0.0843,
+      "step": 12380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.260550866281653e-06,
+      "loss": 0.0554,
+      "step": 12385
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.254997778764994e-06,
+      "loss": 0.0327,
+      "step": 12390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.249444691248334e-06,
+      "loss": 0.0611,
+      "step": 12395
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.243891603731675e-06,
+      "loss": 0.0866,
+      "step": 12400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2383385162150155e-06,
+      "loss": 0.1602,
+      "step": 12405
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2327854286983566e-06,
+      "loss": 0.2472,
+      "step": 12410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.227232341181697e-06,
+      "loss": 0.0588,
+      "step": 12415
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.221679253665039e-06,
+      "loss": 0.0697,
+      "step": 12420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.216126166148379e-06,
+      "loss": 0.0646,
+      "step": 12425
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.21057307863172e-06,
+      "loss": 0.0877,
+      "step": 12430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.20501999111506e-06,
+      "loss": 0.0561,
+      "step": 12435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.199466903598401e-06,
+      "loss": 0.0298,
+      "step": 12440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.193913816081742e-06,
+      "loss": 0.037,
+      "step": 12445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.1883607285650825e-06,
+      "loss": 0.1196,
+      "step": 12450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.182807641048424e-06,
+      "loss": 0.1386,
+      "step": 12455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.177254553531764e-06,
+      "loss": 0.2208,
+      "step": 12460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.171701466015106e-06,
+      "loss": 0.0505,
+      "step": 12465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.166148378498445e-06,
+      "loss": 0.0777,
+      "step": 12470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.160595290981787e-06,
+      "loss": 0.096,
+      "step": 12475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.155042203465127e-06,
+      "loss": 0.0811,
+      "step": 12480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.149489115948468e-06,
+      "loss": 0.0655,
+      "step": 12485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1439360284318085e-06,
+      "loss": 0.0286,
+      "step": 12490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1383829409151496e-06,
+      "loss": 0.0557,
+      "step": 12495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.13282985339849e-06,
+      "loss": 0.0753,
+      "step": 12500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.127276765881831e-06,
+      "loss": 0.1109,
+      "step": 12505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.121723678365171e-06,
+      "loss": 0.1795,
+      "step": 12510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.116170590848512e-06,
+      "loss": 0.0494,
+      "step": 12515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.110617503331852e-06,
+      "loss": 0.0695,
+      "step": 12520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.105064415815194e-06,
+      "loss": 0.07,
+      "step": 12525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.0995113282985344e-06,
+      "loss": 0.0822,
+      "step": 12530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.0939582407818755e-06,
+      "loss": 0.0531,
+      "step": 12535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.089515770768548e-06,
+      "loss": 0.4155,
+      "step": 12540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.083962683251888e-06,
+      "loss": 0.0494,
+      "step": 12545
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.07840959573523e-06,
+      "loss": 0.1038,
+      "step": 12550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.072856508218569e-06,
+      "loss": 0.126,
+      "step": 12555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.067303420701911e-06,
+      "loss": 0.2633,
+      "step": 12560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.0617503331852515e-06,
+      "loss": 0.064,
+      "step": 12565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0561972456685925e-06,
+      "loss": 0.058,
+      "step": 12570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.050644158151933e-06,
+      "loss": 0.0918,
+      "step": 12575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.045091070635274e-06,
+      "loss": 0.1026,
+      "step": 12580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.039537983118614e-06,
+      "loss": 0.0393,
+      "step": 12585
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.033984895601955e-06,
+      "loss": 0.0523,
+      "step": 12590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.028431808085295e-06,
+      "loss": 0.0742,
+      "step": 12595
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.022878720568636e-06,
+      "loss": 0.0861,
+      "step": 12600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.017325633051977e-06,
+      "loss": 0.1553,
+      "step": 12605
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0117725455353185e-06,
+      "loss": 0.279,
+      "step": 12610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.006219458018659e-06,
+      "loss": 0.0604,
+      "step": 12615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.000666370502e-06,
+      "loss": 0.0825,
+      "step": 12620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.99511328298534e-06,
+      "loss": 0.0841,
+      "step": 12625
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.989560195468681e-06,
+      "loss": 0.07,
+      "step": 12630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.984007107952022e-06,
+      "loss": 0.0596,
+      "step": 12635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.978454020435362e-06,
+      "loss": 0.0514,
+      "step": 12640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.972900932918703e-06,
+      "loss": 0.0558,
+      "step": 12645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.967347845402044e-06,
+      "loss": 0.071,
+      "step": 12650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9617947578853855e-06,
+      "loss": 0.1086,
+      "step": 12655
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.956241670368726e-06,
+      "loss": 0.2886,
+      "step": 12660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.950688582852067e-06,
+      "loss": 0.0576,
+      "step": 12665
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.945135495335407e-06,
+      "loss": 0.0563,
+      "step": 12670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.939582407818748e-06,
+      "loss": 0.096,
+      "step": 12675
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.934029320302088e-06,
+      "loss": 0.0969,
+      "step": 12680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.928476232785429e-06,
+      "loss": 0.0355,
+      "step": 12685
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9229231452687696e-06,
+      "loss": 0.0617,
+      "step": 12690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.917370057752111e-06,
+      "loss": 0.0953,
+      "step": 12695
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.911816970235451e-06,
+      "loss": 0.0747,
+      "step": 12700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.906263882718793e-06,
+      "loss": 0.1581,
+      "step": 12705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.900710795202132e-06,
+      "loss": 0.22,
+      "step": 12710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.895157707685474e-06,
+      "loss": 0.054,
+      "step": 12715
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.889604620168814e-06,
+      "loss": 0.0537,
+      "step": 12720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.884051532652155e-06,
+      "loss": 0.0727,
+      "step": 12725
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8784984451354955e-06,
+      "loss": 0.0881,
+      "step": 12730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8729453576188366e-06,
+      "loss": 0.0795,
+      "step": 12735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.867392270102177e-06,
+      "loss": 0.0766,
+      "step": 12740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.861839182585518e-06,
+      "loss": 0.0452,
+      "step": 12745
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.856286095068858e-06,
+      "loss": 0.079,
+      "step": 12750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.850733007552199e-06,
+      "loss": 0.1253,
+      "step": 12755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.845179920035539e-06,
+      "loss": 0.2302,
+      "step": 12760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.839626832518881e-06,
+      "loss": 0.0606,
+      "step": 12765
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.834073745002222e-06,
+      "loss": 0.0561,
+      "step": 12770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8285206574855625e-06,
+      "loss": 0.0763,
+      "step": 12775
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.822967569968904e-06,
+      "loss": 0.0673,
+      "step": 12780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.817414482452244e-06,
+      "loss": 0.1012,
+      "step": 12785
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.811861394935585e-06,
+      "loss": 0.0257,
+      "step": 12790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.806308307418925e-06,
+      "loss": 0.0518,
+      "step": 12795
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.800755219902266e-06,
+      "loss": 0.102,
+      "step": 12800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.795202132385606e-06,
+      "loss": 0.1302,
+      "step": 12805
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.789649044868948e-06,
+      "loss": 0.2529,
+      "step": 12810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.7840959573522885e-06,
+      "loss": 0.0544,
+      "step": 12815
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.7785428698356296e-06,
+      "loss": 0.0657,
+      "step": 12820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.77298978231897e-06,
+      "loss": 0.0862,
+      "step": 12825
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.767436694802311e-06,
+      "loss": 0.0707,
+      "step": 12830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.761883607285651e-06,
+      "loss": 0.0398,
+      "step": 12835
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.756330519768992e-06,
+      "loss": 0.0678,
+      "step": 12840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.750777432252332e-06,
+      "loss": 0.0592,
+      "step": 12845
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.745224344735673e-06,
+      "loss": 0.0633,
+      "step": 12850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.739671257219014e-06,
+      "loss": 0.1238,
+      "step": 12855
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7341181697023555e-06,
+      "loss": 0.2725,
+      "step": 12860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.728565082185695e-06,
+      "loss": 0.0773,
+      "step": 12865
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.723011994669037e-06,
+      "loss": 0.0469,
+      "step": 12870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.717458907152377e-06,
+      "loss": 0.0613,
+      "step": 12875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.711905819635718e-06,
+      "loss": 0.0824,
+      "step": 12880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.706352732119058e-06,
+      "loss": 0.0737,
+      "step": 12885
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.700799644602399e-06,
+      "loss": 0.0439,
+      "step": 12890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.6952465570857396e-06,
+      "loss": 0.061,
+      "step": 12895
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.689693469569081e-06,
+      "loss": 0.0588,
+      "step": 12900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.6841403820524225e-06,
+      "loss": 0.1573,
+      "step": 12905
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.678587294535762e-06,
+      "loss": 0.2104,
+      "step": 12910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.673034207019104e-06,
+      "loss": 0.0515,
+      "step": 12915
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.667481119502444e-06,
+      "loss": 0.0537,
+      "step": 12920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.661928031985785e-06,
+      "loss": 0.0732,
+      "step": 12925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.656374944469125e-06,
+      "loss": 0.0755,
+      "step": 12930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.650821856952466e-06,
+      "loss": 0.0308,
+      "step": 12935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.645268769435807e-06,
+      "loss": 0.0568,
+      "step": 12940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.639715681919148e-06,
+      "loss": 0.0953,
+      "step": 12945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.634162594402488e-06,
+      "loss": 0.0846,
+      "step": 12950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.628609506885829e-06,
+      "loss": 0.1323,
+      "step": 12955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.623056419369169e-06,
+      "loss": 0.1614,
+      "step": 12960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.617503331852511e-06,
+      "loss": 0.0802,
+      "step": 12965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.611950244335851e-06,
+      "loss": 0.0637,
+      "step": 12970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.606397156819192e-06,
+      "loss": 0.0508,
+      "step": 12975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6008440693025325e-06,
+      "loss": 0.0836,
+      "step": 12980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.595290981785874e-06,
+      "loss": 0.0551,
+      "step": 12985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.589737894269214e-06,
+      "loss": 0.1053,
+      "step": 12990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.584184806752555e-06,
+      "loss": 0.0429,
+      "step": 12995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.578631719235895e-06,
+      "loss": 0.0761,
+      "step": 13000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.573078631719236e-06,
+      "loss": 0.1536,
+      "step": 13005
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.567525544202576e-06,
+      "loss": 0.2435,
+      "step": 13010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.561972456685918e-06,
+      "loss": 0.0561,
+      "step": 13015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.556419369169258e-06,
+      "loss": 0.0656,
+      "step": 13020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5508662816525996e-06,
+      "loss": 0.062,
+      "step": 13025
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.54531319413594e-06,
+      "loss": 0.079,
+      "step": 13030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.539760106619281e-06,
+      "loss": 0.067,
+      "step": 13035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.534207019102622e-06,
+      "loss": 0.0296,
+      "step": 13040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.528653931585962e-06,
+      "loss": 0.1216,
+      "step": 13045
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.523100844069303e-06,
+      "loss": 0.0791,
+      "step": 13050
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.517547756552643e-06,
+      "loss": 0.1093,
+      "step": 13055
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.511994669035985e-06,
+      "loss": 0.2942,
+      "step": 13060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.506441581519325e-06,
+      "loss": 0.0634,
+      "step": 13065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5008884940026666e-06,
+      "loss": 0.0771,
+      "step": 13070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.495335406486007e-06,
+      "loss": 0.0598,
+      "step": 13075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.489782318969348e-06,
+      "loss": 0.0798,
+      "step": 13080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.484229231452688e-06,
+      "loss": 0.0405,
+      "step": 13085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.478676143936029e-06,
+      "loss": 0.0699,
+      "step": 13090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.473123056419369e-06,
+      "loss": 0.0763,
+      "step": 13095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.46756996890271e-06,
+      "loss": 0.0934,
+      "step": 13100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.462016881386051e-06,
+      "loss": 0.1859,
+      "step": 13105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.456463793869392e-06,
+      "loss": 0.2356,
+      "step": 13110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.450910706352732e-06,
+      "loss": 0.0738,
+      "step": 13115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.445357618836074e-06,
+      "loss": 0.0646,
+      "step": 13120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.439804531319414e-06,
+      "loss": 0.0638,
+      "step": 13125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.434251443802755e-06,
+      "loss": 0.0923,
+      "step": 13130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.428698356286095e-06,
+      "loss": 0.0363,
+      "step": 13135
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.423145268769436e-06,
+      "loss": 0.0294,
+      "step": 13140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.417592181252777e-06,
+      "loss": 0.0375,
+      "step": 13145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.412039093736118e-06,
+      "loss": 0.0951,
+      "step": 13150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.406486006219458e-06,
+      "loss": 0.1879,
+      "step": 13155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.400932918702799e-06,
+      "loss": 0.1814,
+      "step": 13160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.395379831186139e-06,
+      "loss": 0.0906,
+      "step": 13165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.389826743669481e-06,
+      "loss": 0.0575,
+      "step": 13170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.384273656152822e-06,
+      "loss": 0.0752,
+      "step": 13175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.378720568636162e-06,
+      "loss": 0.0898,
+      "step": 13180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.373167481119503e-06,
+      "loss": 0.0563,
+      "step": 13185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.367614393602844e-06,
+      "loss": 0.0496,
+      "step": 13190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.362061306086185e-06,
+      "loss": 0.0737,
+      "step": 13195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.356508218569525e-06,
+      "loss": 0.1588,
+      "step": 13200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.350955131052866e-06,
+      "loss": 0.1223,
+      "step": 13205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.345402043536206e-06,
+      "loss": 0.2363,
+      "step": 13210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.339848956019548e-06,
+      "loss": 0.0529,
+      "step": 13215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3342958685028874e-06,
+      "loss": 0.0543,
+      "step": 13220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.328742780986229e-06,
+      "loss": 0.0746,
+      "step": 13225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3231896934695696e-06,
+      "loss": 0.0751,
+      "step": 13230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.317636605952911e-06,
+      "loss": 0.0679,
+      "step": 13235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.312083518436251e-06,
+      "loss": 0.0339,
+      "step": 13240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.306530430919592e-06,
+      "loss": 0.1017,
+      "step": 13245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.300977343402932e-06,
+      "loss": 0.0852,
+      "step": 13250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.295424255886273e-06,
+      "loss": 0.1288,
+      "step": 13255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.289871168369613e-06,
+      "loss": 0.2528,
+      "step": 13260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2843180808529545e-06,
+      "loss": 0.0578,
+      "step": 13265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.278764993336295e-06,
+      "loss": 0.0495,
+      "step": 13270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2732119058196366e-06,
+      "loss": 0.0798,
+      "step": 13275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.267658818302977e-06,
+      "loss": 0.0888,
+      "step": 13280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.262105730786318e-06,
+      "loss": 0.0335,
+      "step": 13285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.256552643269658e-06,
+      "loss": 0.0405,
+      "step": 13290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.250999555752999e-06,
+      "loss": 0.073,
+      "step": 13295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.245446468236339e-06,
+      "loss": 0.1129,
+      "step": 13300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.23989338071968e-06,
+      "loss": 0.1393,
+      "step": 13305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.2343402932030215e-06,
+      "loss": 0.2412,
+      "step": 13310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.228787205686362e-06,
+      "loss": 0.0574,
+      "step": 13315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.223234118169704e-06,
+      "loss": 0.0687,
+      "step": 13320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.217681030653044e-06,
+      "loss": 0.0642,
+      "step": 13325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.212127943136385e-06,
+      "loss": 0.0731,
+      "step": 13330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.206574855619725e-06,
+      "loss": 0.0379,
+      "step": 13335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.201021768103066e-06,
+      "loss": 0.0405,
+      "step": 13340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.195468680586406e-06,
+      "loss": 0.0416,
+      "step": 13345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.1899155930697474e-06,
+      "loss": 0.0693,
+      "step": 13350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.184362505553088e-06,
+      "loss": 0.1614,
+      "step": 13355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.178809418036429e-06,
+      "loss": 0.2359,
+      "step": 13360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.173256330519769e-06,
+      "loss": 0.049,
+      "step": 13365
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.167703243003111e-06,
+      "loss": 0.0714,
+      "step": 13370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.16215015548645e-06,
+      "loss": 0.0754,
+      "step": 13375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.156597067969792e-06,
+      "loss": 0.1066,
+      "step": 13380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.151043980453132e-06,
+      "loss": 0.0325,
+      "step": 13385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.145490892936473e-06,
+      "loss": 0.0861,
+      "step": 13390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.139937805419814e-06,
+      "loss": 0.0466,
+      "step": 13395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.134384717903155e-06,
+      "loss": 0.132,
+      "step": 13400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.128831630386495e-06,
+      "loss": 0.1693,
+      "step": 13405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.123278542869836e-06,
+      "loss": 0.2495,
+      "step": 13410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.117725455353176e-06,
+      "loss": 0.0579,
+      "step": 13415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.112172367836517e-06,
+      "loss": 0.0703,
+      "step": 13420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.1066192803198574e-06,
+      "loss": 0.1037,
+      "step": 13425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.101066192803199e-06,
+      "loss": 0.0837,
+      "step": 13430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.0955131052865396e-06,
+      "loss": 0.0601,
+      "step": 13435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.089960017769881e-06,
+      "loss": 0.0532,
+      "step": 13440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.084406930253222e-06,
+      "loss": 0.0667,
+      "step": 13445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.078853842736562e-06,
+      "loss": 0.0986,
+      "step": 13450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.073300755219903e-06,
+      "loss": 0.1116,
+      "step": 13455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.067747667703243e-06,
+      "loss": 0.2368,
+      "step": 13460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.062194580186584e-06,
+      "loss": 0.0586,
+      "step": 13465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0566414926699245e-06,
+      "loss": 0.0654,
+      "step": 13470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.051088405153266e-06,
+      "loss": 0.0871,
+      "step": 13475
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.045535317636607e-06,
+      "loss": 0.0598,
+      "step": 13480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.039982230119948e-06,
+      "loss": 0.065,
+      "step": 13485
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.034429142603288e-06,
+      "loss": 0.0366,
+      "step": 13490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.028876055086629e-06,
+      "loss": 0.08,
+      "step": 13495
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.023322967569969e-06,
+      "loss": 0.0897,
+      "step": 13500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.01776988005331e-06,
+      "loss": 0.1548,
+      "step": 13505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.01221679253665e-06,
+      "loss": 0.3389,
+      "step": 13510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0066637050199915e-06,
+      "loss": 0.0602,
+      "step": 13515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.001110617503332e-06,
+      "loss": 0.0744,
+      "step": 13520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.995557529986674e-06,
+      "loss": 0.0692,
+      "step": 13525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.990004442470014e-06,
+      "loss": 0.0759,
+      "step": 13530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.984451354953355e-06,
+      "loss": 0.0509,
+      "step": 13535
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.978898267436695e-06,
+      "loss": 0.0206,
+      "step": 13540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.973345179920036e-06,
+      "loss": 0.0467,
+      "step": 13545
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.967792092403376e-06,
+      "loss": 0.0769,
+      "step": 13550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9622390048867174e-06,
+      "loss": 0.1698,
+      "step": 13555
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9566859173700585e-06,
+      "loss": 0.227,
+      "step": 13560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.951132829853399e-06,
+      "loss": 0.0708,
+      "step": 13565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.94557974233674e-06,
+      "loss": 0.0702,
+      "step": 13570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.94002665482008e-06,
+      "loss": 0.0766,
+      "step": 13575
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.934473567303421e-06,
+      "loss": 0.0652,
+      "step": 13580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.928920479786762e-06,
+      "loss": 0.0571,
+      "step": 13585
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.923367392270102e-06,
+      "loss": 0.0648,
+      "step": 13590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.917814304753443e-06,
+      "loss": 0.0454,
+      "step": 13595
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.912261217236784e-06,
+      "loss": 0.1342,
+      "step": 13600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.906708129720125e-06,
+      "loss": 0.1654,
+      "step": 13605
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.901155042203466e-06,
+      "loss": 0.2228,
+      "step": 13610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.895601954686806e-06,
+      "loss": 0.0578,
+      "step": 13615
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.890048867170147e-06,
+      "loss": 0.0548,
+      "step": 13620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.884495779653487e-06,
+      "loss": 0.0941,
+      "step": 13625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.878942692136828e-06,
+      "loss": 0.0875,
+      "step": 13630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.873389604620169e-06,
+      "loss": 0.0203,
+      "step": 13635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.8678365171035096e-06,
+      "loss": 0.0259,
+      "step": 13640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.862283429586851e-06,
+      "loss": 0.0657,
+      "step": 13645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.856730342070192e-06,
+      "loss": 0.0903,
+      "step": 13650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.851177254553533e-06,
+      "loss": 0.0966,
+      "step": 13655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.845624167036873e-06,
+      "loss": 0.2744,
+      "step": 13660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.840071079520214e-06,
+      "loss": 0.0586,
+      "step": 13665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.834517992003554e-06,
+      "loss": 0.0759,
+      "step": 13670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.828964904486895e-06,
+      "loss": 0.0834,
+      "step": 13675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.823411816970236e-06,
+      "loss": 0.0909,
+      "step": 13680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.817858729453577e-06,
+      "loss": 0.0378,
+      "step": 13685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.812305641936918e-06,
+      "loss": 0.0379,
+      "step": 13690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.806752554420258e-06,
+      "loss": 0.0712,
+      "step": 13695
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.801199466903599e-06,
+      "loss": 0.1161,
+      "step": 13700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.795646379386939e-06,
+      "loss": 0.1397,
+      "step": 13705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.79009329187028e-06,
+      "loss": 0.3017,
+      "step": 13710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.784540204353621e-06,
+      "loss": 0.0687,
+      "step": 13715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7789871168369615e-06,
+      "loss": 0.0664,
+      "step": 13720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7734340293203025e-06,
+      "loss": 0.0766,
+      "step": 13725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.767880941803643e-06,
+      "loss": 0.0785,
+      "step": 13730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.762327854286984e-06,
+      "loss": 0.0572,
+      "step": 13735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.756774766770325e-06,
+      "loss": 0.0619,
+      "step": 13740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.751221679253665e-06,
+      "loss": 0.0543,
+      "step": 13745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745668591737006e-06,
+      "loss": 0.1095,
+      "step": 13750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.740115504220346e-06,
+      "loss": 0.1272,
+      "step": 13755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7345624167036874e-06,
+      "loss": 0.2245,
+      "step": 13760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7290093291870285e-06,
+      "loss": 0.0494,
+      "step": 13765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.723456241670369e-06,
+      "loss": 0.054,
+      "step": 13770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.71790315415371e-06,
+      "loss": 0.0703,
+      "step": 13775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.712350066637051e-06,
+      "loss": 0.0677,
+      "step": 13780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.706796979120392e-06,
+      "loss": 0.0437,
+      "step": 13785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.701243891603732e-06,
+      "loss": 0.0208,
+      "step": 13790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.695690804087073e-06,
+      "loss": 0.0511,
+      "step": 13795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.690137716570413e-06,
+      "loss": 0.0624,
+      "step": 13800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6845846290537545e-06,
+      "loss": 0.1252,
+      "step": 13805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.6790315415370955e-06,
+      "loss": 0.2626,
+      "step": 13810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.673478454020436e-06,
+      "loss": 0.0563,
+      "step": 13815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.667925366503777e-06,
+      "loss": 0.0615,
+      "step": 13820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.662372278987117e-06,
+      "loss": 0.0605,
+      "step": 13825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.656819191470458e-06,
+      "loss": 0.0981,
+      "step": 13830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.651266103953799e-06,
+      "loss": 0.0391,
+      "step": 13835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.645713016437139e-06,
+      "loss": 0.0593,
+      "step": 13840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.64015992892048e-06,
+      "loss": 0.0835,
+      "step": 13845
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.634606841403821e-06,
+      "loss": 0.0966,
+      "step": 13850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.629053753887162e-06,
+      "loss": 0.1804,
+      "step": 13855
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.623500666370502e-06,
+      "loss": 0.2938,
+      "step": 13860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.617947578853843e-06,
+      "loss": 0.061,
+      "step": 13865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.612394491337184e-06,
+      "loss": 0.0798,
+      "step": 13870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.606841403820524e-06,
+      "loss": 0.0693,
+      "step": 13875
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.601288316303865e-06,
+      "loss": 0.0772,
+      "step": 13880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5957352287872055e-06,
+      "loss": 0.0576,
+      "step": 13885
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.590182141270547e-06,
+      "loss": 0.0567,
+      "step": 13890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.584629053753888e-06,
+      "loss": 0.0605,
+      "step": 13895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.579075966237228e-06,
+      "loss": 0.0796,
+      "step": 13900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.573522878720569e-06,
+      "loss": 0.1112,
+      "step": 13905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.567969791203909e-06,
+      "loss": 0.248,
+      "step": 13910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.562416703687251e-06,
+      "loss": 0.0721,
+      "step": 13915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.556863616170591e-06,
+      "loss": 0.0514,
+      "step": 13920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.551310528653932e-06,
+      "loss": 0.0729,
+      "step": 13925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5457574411372725e-06,
+      "loss": 0.1085,
+      "step": 13930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.540204353620614e-06,
+      "loss": 0.0554,
+      "step": 13935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.534651266103955e-06,
+      "loss": 0.0583,
+      "step": 13940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.529098178587295e-06,
+      "loss": 0.0531,
+      "step": 13945
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.523545091070636e-06,
+      "loss": 0.1009,
+      "step": 13950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.517992003553976e-06,
+      "loss": 0.1851,
+      "step": 13955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.512438916037317e-06,
+      "loss": 0.2653,
+      "step": 13960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.506885828520658e-06,
+      "loss": 0.048,
+      "step": 13965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5013327410039985e-06,
+      "loss": 0.0815,
+      "step": 13970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4957796534873396e-06,
+      "loss": 0.0778,
+      "step": 13975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.49022656597068e-06,
+      "loss": 0.0635,
+      "step": 13980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.484673478454021e-06,
+      "loss": 0.0347,
+      "step": 13985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.479120390937362e-06,
+      "loss": 0.0572,
+      "step": 13990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.473567303420702e-06,
+      "loss": 0.0781,
+      "step": 13995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.468014215904043e-06,
+      "loss": 0.1064,
+      "step": 14000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 0.02988557517528534,
+      "eval_runtime": 1484.3586,
+      "eval_samples_per_second": 10.791,
+      "eval_steps_per_second": 2.698,
+      "eval_wer": 0.0711503786512802,
+      "step": 14000
+    }
+  ],
+  "max_steps": 18018,
+  "num_train_epochs": 2,
+  "total_flos": 6.942003676372472e+18,
+  "trial_name": null,
+  "trial_params": null
+}