diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4688 @@
+{
+  "best_metric": 39.565217391304344,
+  "best_model_checkpoint": "./whisper-small-wer35-ekg\\checkpoint-6000",
+  "epoch": 95.58823529411765,
+  "eval_steps": 500,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.15,
+      "grad_norm": 28.624893188476562,
+      "learning_rate": 9.433962264150944e-08,
+      "loss": 2.7579,
+      "step": 10
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 28.752416610717773,
+      "learning_rate": 2.1226415094339622e-07,
+      "loss": 2.6811,
+      "step": 20
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 25.40899658203125,
+      "learning_rate": 3.3018867924528305e-07,
+      "loss": 2.5367,
+      "step": 30
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 22.991146087646484,
+      "learning_rate": 4.4811320754716983e-07,
+      "loss": 2.2512,
+      "step": 40
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 18.11638832092285,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 2.0029,
+      "step": 50
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 17.4566650390625,
+      "learning_rate": 6.839622641509434e-07,
+      "loss": 1.7014,
+      "step": 60
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 12.19814395904541,
+      "learning_rate": 8.018867924528302e-07,
+      "loss": 1.4435,
+      "step": 70
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 9.45124626159668,
+      "learning_rate": 9.19811320754717e-07,
+      "loss": 1.2141,
+      "step": 80
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 8.163089752197266,
+      "learning_rate": 1.037735849056604e-06,
+      "loss": 1.0621,
+      "step": 90
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 6.771365642547607,
+      "learning_rate": 1.1556603773584908e-06,
+      "loss": 0.9325,
+      "step": 100
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 5.819299697875977,
+      "learning_rate": 1.2735849056603775e-06,
+      "loss": 0.8767,
+      "step": 110
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 5.547601222991943,
+      "learning_rate": 1.3915094339622643e-06,
+      "loss": 0.7872,
+      "step": 120
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 4.7814154624938965,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 0.7148,
+      "step": 130
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 4.755609512329102,
+      "learning_rate": 1.6273584905660379e-06,
+      "loss": 0.6742,
+      "step": 140
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 4.354093074798584,
+      "learning_rate": 1.7452830188679247e-06,
+      "loss": 0.6191,
+      "step": 150
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 4.038461685180664,
+      "learning_rate": 1.8632075471698114e-06,
+      "loss": 0.5673,
+      "step": 160
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 4.2933526039123535,
+      "learning_rate": 1.981132075471698e-06,
+      "loss": 0.5534,
+      "step": 170
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 4.423107147216797,
+      "learning_rate": 2.099056603773585e-06,
+      "loss": 0.5162,
+      "step": 180
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 3.7280497550964355,
+      "learning_rate": 2.2169811320754718e-06,
+      "loss": 0.4796,
+      "step": 190
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 3.990518808364868,
+      "learning_rate": 2.3349056603773588e-06,
+      "loss": 0.4976,
+      "step": 200
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 3.825350046157837,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 0.4493,
+      "step": 210
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 4.116565704345703,
+      "learning_rate": 2.5707547169811327e-06,
+      "loss": 0.429,
+      "step": 220
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 3.6445629596710205,
+      "learning_rate": 2.688679245283019e-06,
+      "loss": 0.3898,
+      "step": 230
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 3.741975784301758,
+      "learning_rate": 2.8066037735849063e-06,
+      "loss": 0.394,
+      "step": 240
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 3.4814412593841553,
+      "learning_rate": 2.9245283018867924e-06,
+      "loss": 0.4115,
+      "step": 250
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 3.914616823196411,
+      "learning_rate": 3.04245283018868e-06,
+      "loss": 0.3858,
+      "step": 260
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 3.9170775413513184,
+      "learning_rate": 3.160377358490566e-06,
+      "loss": 0.3731,
+      "step": 270
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 3.542541980743408,
+      "learning_rate": 3.2783018867924534e-06,
+      "loss": 0.347,
+      "step": 280
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.5728800296783447,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 0.3171,
+      "step": 290
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 3.627119541168213,
+      "learning_rate": 3.514150943396227e-06,
+      "loss": 0.3105,
+      "step": 300
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 3.763631820678711,
+      "learning_rate": 3.632075471698113e-06,
+      "loss": 0.315,
+      "step": 310
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 3.1514198780059814,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.3116,
+      "step": 320
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 3.847184181213379,
+      "learning_rate": 3.8679245283018875e-06,
+      "loss": 0.3076,
+      "step": 330
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 3.8552136421203613,
+      "learning_rate": 3.985849056603774e-06,
+      "loss": 0.3106,
+      "step": 340
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 3.341456174850464,
+      "learning_rate": 4.103773584905661e-06,
+      "loss": 0.2699,
+      "step": 350
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 3.3753416538238525,
+      "learning_rate": 4.221698113207548e-06,
+      "loss": 0.2442,
+      "step": 360
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 3.4466023445129395,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 0.2436,
+      "step": 370
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 3.2262954711914062,
+      "learning_rate": 4.457547169811321e-06,
+      "loss": 0.2524,
+      "step": 380
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 3.167746067047119,
+      "learning_rate": 4.575471698113208e-06,
+      "loss": 0.2551,
+      "step": 390
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 3.338438034057617,
+      "learning_rate": 4.693396226415095e-06,
+      "loss": 0.2525,
+      "step": 400
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 2.4594788551330566,
+      "learning_rate": 4.811320754716982e-06,
+      "loss": 0.2243,
+      "step": 410
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 3.1020987033843994,
+      "learning_rate": 4.929245283018868e-06,
+      "loss": 0.2055,
+      "step": 420
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 2.919455051422119,
+      "learning_rate": 5.047169811320756e-06,
+      "loss": 0.2107,
+      "step": 430
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 3.4662413597106934,
+      "learning_rate": 5.165094339622642e-06,
+      "loss": 0.1946,
+      "step": 440
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 3.209831476211548,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 0.2049,
+      "step": 450
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 3.5380163192749023,
+      "learning_rate": 5.400943396226416e-06,
+      "loss": 0.1865,
+      "step": 460
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 3.4075424671173096,
+      "learning_rate": 5.518867924528303e-06,
+      "loss": 0.184,
+      "step": 470
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 2.4087564945220947,
+      "learning_rate": 5.636792452830189e-06,
+      "loss": 0.1725,
+      "step": 480
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 3.172959566116333,
+      "learning_rate": 5.754716981132076e-06,
+      "loss": 0.1505,
+      "step": 490
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 3.0081095695495605,
+      "learning_rate": 5.872641509433963e-06,
+      "loss": 0.149,
+      "step": 500
+    },
+    {
+      "epoch": 7.35,
+      "eval_loss": 0.32423341274261475,
+      "eval_runtime": 476.393,
+      "eval_samples_per_second": 1.583,
+      "eval_steps_per_second": 0.017,
+      "eval_wer": 50.19762845849802,
+      "step": 500
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 2.85640811920166,
+      "learning_rate": 5.99056603773585e-06,
+      "loss": 0.1478,
+      "step": 510
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 3.185556650161743,
+      "learning_rate": 6.108490566037736e-06,
+      "loss": 0.1495,
+      "step": 520
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 2.893033266067505,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 0.1508,
+      "step": 530
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 3.044069290161133,
+      "learning_rate": 6.34433962264151e-06,
+      "loss": 0.155,
+      "step": 540
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 2.3645215034484863,
+      "learning_rate": 6.462264150943397e-06,
+      "loss": 0.1191,
+      "step": 550
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 2.373063564300537,
+      "learning_rate": 6.580188679245284e-06,
+      "loss": 0.1092,
+      "step": 560
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 2.313464641571045,
+      "learning_rate": 6.69811320754717e-06,
+      "loss": 0.1108,
+      "step": 570
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 2.749617099761963,
+      "learning_rate": 6.816037735849057e-06,
+      "loss": 0.108,
+      "step": 580
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 2.794914960861206,
+      "learning_rate": 6.933962264150944e-06,
+      "loss": 0.1102,
+      "step": 590
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 2.7622482776641846,
+      "learning_rate": 7.051886792452831e-06,
+      "loss": 0.1121,
+      "step": 600
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 2.6082873344421387,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 0.1171,
+      "step": 610
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 2.379115581512451,
+      "learning_rate": 7.287735849056604e-06,
+      "loss": 0.0824,
+      "step": 620
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 3.37858510017395,
+      "learning_rate": 7.405660377358491e-06,
+      "loss": 0.0834,
+      "step": 630
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 2.881742000579834,
+      "learning_rate": 7.523584905660378e-06,
+      "loss": 0.0809,
+      "step": 640
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 2.6887032985687256,
+      "learning_rate": 7.641509433962266e-06,
+      "loss": 0.0832,
+      "step": 650
+    },
+    {
+      "epoch": 9.71,
+      "grad_norm": 2.685237407684326,
+      "learning_rate": 7.75943396226415e-06,
+      "loss": 0.0815,
+      "step": 660
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 2.297445774078369,
+      "learning_rate": 7.877358490566038e-06,
+      "loss": 0.078,
+      "step": 670
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 3.055098056793213,
+      "learning_rate": 7.995283018867925e-06,
+      "loss": 0.0753,
+      "step": 680
+    },
+    {
+      "epoch": 10.15,
+      "grad_norm": 2.498236656188965,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 0.054,
+      "step": 690
+    },
+    {
+      "epoch": 10.29,
+      "grad_norm": 2.1496338844299316,
+      "learning_rate": 8.231132075471699e-06,
+      "loss": 0.0583,
+      "step": 700
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 2.9474973678588867,
+      "learning_rate": 8.349056603773585e-06,
+      "loss": 0.0559,
+      "step": 710
+    },
+    {
+      "epoch": 10.59,
+      "grad_norm": 3.049436569213867,
+      "learning_rate": 8.466981132075472e-06,
+      "loss": 0.0557,
+      "step": 720
+    },
+    {
+      "epoch": 10.74,
+      "grad_norm": 2.463704824447632,
+      "learning_rate": 8.58490566037736e-06,
+      "loss": 0.058,
+      "step": 730
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 2.369274139404297,
+      "learning_rate": 8.702830188679245e-06,
+      "loss": 0.0614,
+      "step": 740
+    },
+    {
+      "epoch": 11.03,
+      "grad_norm": 2.1053879261016846,
+      "learning_rate": 8.820754716981133e-06,
+      "loss": 0.0545,
+      "step": 750
+    },
+    {
+      "epoch": 11.18,
+      "grad_norm": 1.6645501852035522,
+      "learning_rate": 8.938679245283019e-06,
+      "loss": 0.0404,
+      "step": 760
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 2.5582962036132812,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 0.0407,
+      "step": 770
+    },
+    {
+      "epoch": 11.47,
+      "grad_norm": 2.079361915588379,
+      "learning_rate": 9.174528301886794e-06,
+      "loss": 0.0392,
+      "step": 780
+    },
+    {
+      "epoch": 11.62,
+      "grad_norm": 2.74023699760437,
+      "learning_rate": 9.292452830188679e-06,
+      "loss": 0.0437,
+      "step": 790
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 2.8623881340026855,
+      "learning_rate": 9.410377358490567e-06,
+      "loss": 0.0416,
+      "step": 800
+    },
+    {
+      "epoch": 11.91,
+      "grad_norm": 2.3341763019561768,
+      "learning_rate": 9.528301886792455e-06,
+      "loss": 0.0435,
+      "step": 810
+    },
+    {
+      "epoch": 12.06,
+      "grad_norm": 2.0387160778045654,
+      "learning_rate": 9.64622641509434e-06,
+      "loss": 0.04,
+      "step": 820
+    },
+    {
+      "epoch": 12.21,
+      "grad_norm": 2.4834096431732178,
+      "learning_rate": 9.764150943396227e-06,
+      "loss": 0.029,
+      "step": 830
+    },
+    {
+      "epoch": 12.35,
+      "grad_norm": 2.721518039703369,
+      "learning_rate": 9.882075471698113e-06,
+      "loss": 0.0297,
+      "step": 840
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 2.684687376022339,
+      "learning_rate": 1e-05,
+      "loss": 0.0316,
+      "step": 850
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 2.5442607402801514,
+      "learning_rate": 9.983153638814017e-06,
+      "loss": 0.031,
+      "step": 860
+    },
+    {
+      "epoch": 12.79,
+      "grad_norm": 3.274472236633301,
+      "learning_rate": 9.966307277628034e-06,
+      "loss": 0.0328,
+      "step": 870
+    },
+    {
+      "epoch": 12.94,
+      "grad_norm": 2.6189253330230713,
+      "learning_rate": 9.94946091644205e-06,
+      "loss": 0.0328,
+      "step": 880
+    },
+    {
+      "epoch": 13.09,
+      "grad_norm": 2.8672521114349365,
+      "learning_rate": 9.932614555256066e-06,
+      "loss": 0.0299,
+      "step": 890
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 2.2015247344970703,
+      "learning_rate": 9.915768194070081e-06,
+      "loss": 0.0224,
+      "step": 900
+    },
+    {
+      "epoch": 13.38,
+      "grad_norm": 1.7655293941497803,
+      "learning_rate": 9.898921832884097e-06,
+      "loss": 0.0222,
+      "step": 910
+    },
+    {
+      "epoch": 13.53,
+      "grad_norm": 2.5391361713409424,
+      "learning_rate": 9.882075471698113e-06,
+      "loss": 0.0232,
+      "step": 920
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 2.2108376026153564,
+      "learning_rate": 9.86522911051213e-06,
+      "loss": 0.0249,
+      "step": 930
+    },
+    {
+      "epoch": 13.82,
+      "grad_norm": 2.0707106590270996,
+      "learning_rate": 9.848382749326146e-06,
+      "loss": 0.0241,
+      "step": 940
+    },
+    {
+      "epoch": 13.97,
+      "grad_norm": 2.2643473148345947,
+      "learning_rate": 9.831536388140162e-06,
+      "loss": 0.0276,
+      "step": 950
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 1.739284634590149,
+      "learning_rate": 9.81469002695418e-06,
+      "loss": 0.017,
+      "step": 960
+    },
+    {
+      "epoch": 14.26,
+      "grad_norm": 1.8066262006759644,
+      "learning_rate": 9.797843665768195e-06,
+      "loss": 0.0157,
+      "step": 970
+    },
+    {
+      "epoch": 14.41,
+      "grad_norm": 2.5150985717773438,
+      "learning_rate": 9.780997304582211e-06,
+      "loss": 0.0196,
+      "step": 980
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 1.7605470418930054,
+      "learning_rate": 9.764150943396227e-06,
+      "loss": 0.0169,
+      "step": 990
+    },
+    {
+      "epoch": 14.71,
+      "grad_norm": 2.1510426998138428,
+      "learning_rate": 9.747304582210243e-06,
+      "loss": 0.0186,
+      "step": 1000
+    },
+    {
+      "epoch": 14.71,
+      "eval_loss": 0.4280255138874054,
+      "eval_runtime": 408.4481,
+      "eval_samples_per_second": 1.846,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 46.126482213438734,
+      "step": 1000
+    },
+    {
+      "epoch": 14.85,
+      "grad_norm": 2.3640592098236084,
+      "learning_rate": 9.73045822102426e-06,
+      "loss": 0.0187,
+      "step": 1010
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 1.8428599834442139,
+      "learning_rate": 9.713611859838276e-06,
+      "loss": 0.0213,
+      "step": 1020
+    },
+    {
+      "epoch": 15.15,
+      "grad_norm": 1.8431464433670044,
+      "learning_rate": 9.696765498652292e-06,
+      "loss": 0.0127,
+      "step": 1030
+    },
+    {
+      "epoch": 15.29,
+      "grad_norm": 1.7996199131011963,
+      "learning_rate": 9.679919137466307e-06,
+      "loss": 0.0137,
+      "step": 1040
+    },
+    {
+      "epoch": 15.44,
+      "grad_norm": 1.7150269746780396,
+      "learning_rate": 9.663072776280325e-06,
+      "loss": 0.0148,
+      "step": 1050
+    },
+    {
+      "epoch": 15.59,
+      "grad_norm": 1.8035845756530762,
+      "learning_rate": 9.64622641509434e-06,
+      "loss": 0.0137,
+      "step": 1060
+    },
+    {
+      "epoch": 15.74,
+      "grad_norm": 2.240055799484253,
+      "learning_rate": 9.629380053908356e-06,
+      "loss": 0.0146,
+      "step": 1070
+    },
+    {
+      "epoch": 15.88,
+      "grad_norm": 2.2169549465179443,
+      "learning_rate": 9.612533692722372e-06,
+      "loss": 0.0129,
+      "step": 1080
+    },
+    {
+      "epoch": 16.03,
+      "grad_norm": 1.7007489204406738,
+      "learning_rate": 9.595687331536388e-06,
+      "loss": 0.0143,
+      "step": 1090
+    },
+    {
+      "epoch": 16.18,
+      "grad_norm": 1.3945379257202148,
+      "learning_rate": 9.578840970350406e-06,
+      "loss": 0.0111,
+      "step": 1100
+    },
+    {
+      "epoch": 16.32,
+      "grad_norm": 1.6498446464538574,
+      "learning_rate": 9.561994609164421e-06,
+      "loss": 0.0108,
+      "step": 1110
+    },
+    {
+      "epoch": 16.47,
+      "grad_norm": 1.7959011793136597,
+      "learning_rate": 9.545148247978437e-06,
+      "loss": 0.012,
+      "step": 1120
+    },
+    {
+      "epoch": 16.62,
+      "grad_norm": 1.7820062637329102,
+      "learning_rate": 9.528301886792455e-06,
+      "loss": 0.0133,
+      "step": 1130
+    },
+    {
+      "epoch": 16.76,
+      "grad_norm": 1.8330453634262085,
+      "learning_rate": 9.51145552560647e-06,
+      "loss": 0.0134,
+      "step": 1140
+    },
+    {
+      "epoch": 16.91,
+      "grad_norm": 1.6996911764144897,
+      "learning_rate": 9.494609164420486e-06,
+      "loss": 0.0121,
+      "step": 1150
+    },
+    {
+      "epoch": 17.06,
+      "grad_norm": 1.2411192655563354,
+      "learning_rate": 9.477762803234502e-06,
+      "loss": 0.0112,
+      "step": 1160
+    },
+    {
+      "epoch": 17.21,
+      "grad_norm": 1.4605441093444824,
+      "learning_rate": 9.460916442048518e-06,
+      "loss": 0.0108,
+      "step": 1170
+    },
+    {
+      "epoch": 17.35,
+      "grad_norm": 1.6172523498535156,
+      "learning_rate": 9.444070080862533e-06,
+      "loss": 0.0093,
+      "step": 1180
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 1.367756724357605,
+      "learning_rate": 9.427223719676551e-06,
+      "loss": 0.0124,
+      "step": 1190
+    },
+    {
+      "epoch": 17.65,
+      "grad_norm": 2.769695997238159,
+      "learning_rate": 9.410377358490567e-06,
+      "loss": 0.0111,
+      "step": 1200
+    },
+    {
+      "epoch": 17.79,
+      "grad_norm": 1.4088915586471558,
+      "learning_rate": 9.393530997304582e-06,
+      "loss": 0.0118,
+      "step": 1210
+    },
+    {
+      "epoch": 17.94,
+      "grad_norm": 2.075134038925171,
+      "learning_rate": 9.3766846361186e-06,
+      "loss": 0.0113,
+      "step": 1220
+    },
+    {
+      "epoch": 18.09,
+      "grad_norm": 1.1849188804626465,
+      "learning_rate": 9.359838274932616e-06,
+      "loss": 0.0095,
+      "step": 1230
+    },
+    {
+      "epoch": 18.24,
+      "grad_norm": 0.7864483594894409,
+      "learning_rate": 9.342991913746632e-06,
+      "loss": 0.008,
+      "step": 1240
+    },
+    {
+      "epoch": 18.38,
+      "grad_norm": 1.687619924545288,
+      "learning_rate": 9.326145552560647e-06,
+      "loss": 0.0082,
+      "step": 1250
+    },
+    {
+      "epoch": 18.53,
+      "grad_norm": 2.207996368408203,
+      "learning_rate": 9.309299191374663e-06,
+      "loss": 0.0105,
+      "step": 1260
+    },
+    {
+      "epoch": 18.68,
+      "grad_norm": 1.876765489578247,
+      "learning_rate": 9.292452830188679e-06,
+      "loss": 0.009,
+      "step": 1270
+    },
+    {
+      "epoch": 18.82,
+      "grad_norm": 2.1015830039978027,
+      "learning_rate": 9.275606469002696e-06,
+      "loss": 0.0113,
+      "step": 1280
+    },
+    {
+      "epoch": 18.97,
+      "grad_norm": 1.7208313941955566,
+      "learning_rate": 9.258760107816712e-06,
+      "loss": 0.0116,
+      "step": 1290
+    },
+    {
+      "epoch": 19.12,
+      "grad_norm": 1.086099624633789,
+      "learning_rate": 9.241913746630728e-06,
+      "loss": 0.0085,
+      "step": 1300
+    },
+    {
+      "epoch": 19.26,
+      "grad_norm": 1.8027433156967163,
+      "learning_rate": 9.225067385444745e-06,
+      "loss": 0.0088,
+      "step": 1310
+    },
+    {
+      "epoch": 19.41,
+      "grad_norm": 1.5787283182144165,
+      "learning_rate": 9.208221024258761e-06,
+      "loss": 0.0075,
+      "step": 1320
+    },
+    {
+      "epoch": 19.56,
+      "grad_norm": 1.3783093690872192,
+      "learning_rate": 9.191374663072777e-06,
+      "loss": 0.0086,
+      "step": 1330
+    },
+    {
+      "epoch": 19.71,
+      "grad_norm": 1.2076382637023926,
+      "learning_rate": 9.174528301886794e-06,
+      "loss": 0.0084,
+      "step": 1340
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 1.688698649406433,
+      "learning_rate": 9.157681940700809e-06,
+      "loss": 0.008,
+      "step": 1350
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 1.9429757595062256,
+      "learning_rate": 9.140835579514824e-06,
+      "loss": 0.0081,
+      "step": 1360
+    },
+    {
+      "epoch": 20.15,
+      "grad_norm": 1.4460784196853638,
+      "learning_rate": 9.123989218328842e-06,
+      "loss": 0.0067,
+      "step": 1370
+    },
+    {
+      "epoch": 20.29,
+      "grad_norm": 1.672726035118103,
+      "learning_rate": 9.107142857142858e-06,
+      "loss": 0.0068,
+      "step": 1380
+    },
+    {
+      "epoch": 20.44,
+      "grad_norm": 0.6751158833503723,
+      "learning_rate": 9.090296495956873e-06,
+      "loss": 0.0066,
+      "step": 1390
+    },
+    {
+      "epoch": 20.59,
+      "grad_norm": 1.696090817451477,
+      "learning_rate": 9.07345013477089e-06,
+      "loss": 0.0066,
+      "step": 1400
+    },
+    {
+      "epoch": 20.74,
+      "grad_norm": 1.0858286619186401,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 0.0065,
+      "step": 1410
+    },
+    {
+      "epoch": 20.88,
+      "grad_norm": 1.6216528415679932,
+      "learning_rate": 9.039757412398922e-06,
+      "loss": 0.0069,
+      "step": 1420
+    },
+    {
+      "epoch": 21.03,
+      "grad_norm": 1.192878246307373,
+      "learning_rate": 9.02291105121294e-06,
+      "loss": 0.0063,
+      "step": 1430
+    },
+    {
+      "epoch": 21.18,
+      "grad_norm": 1.2905162572860718,
+      "learning_rate": 9.006064690026954e-06,
+      "loss": 0.0062,
+      "step": 1440
+    },
+    {
+      "epoch": 21.32,
+      "grad_norm": 1.0860906839370728,
+      "learning_rate": 8.989218328840971e-06,
+      "loss": 0.0058,
+      "step": 1450
+    },
+    {
+      "epoch": 21.47,
+      "grad_norm": 1.6400948762893677,
+      "learning_rate": 8.972371967654987e-06,
+      "loss": 0.0066,
+      "step": 1460
+    },
+    {
+      "epoch": 21.62,
+      "grad_norm": 2.1465609073638916,
+      "learning_rate": 8.955525606469003e-06,
+      "loss": 0.0097,
+      "step": 1470
+    },
+    {
+      "epoch": 21.76,
+      "grad_norm": 0.9771747589111328,
+      "learning_rate": 8.938679245283019e-06,
+      "loss": 0.0074,
+      "step": 1480
+    },
+    {
+      "epoch": 21.91,
+      "grad_norm": 1.7411590814590454,
+      "learning_rate": 8.921832884097036e-06,
+      "loss": 0.0068,
+      "step": 1490
+    },
+    {
+      "epoch": 22.06,
+      "grad_norm": 1.491974949836731,
+      "learning_rate": 8.904986522911052e-06,
+      "loss": 0.006,
+      "step": 1500
+    },
+    {
+      "epoch": 22.06,
+      "eval_loss": 0.44684624671936035,
+      "eval_runtime": 395.7518,
+      "eval_samples_per_second": 1.905,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 43.537549407114625,
+      "step": 1500
+    },
+    {
+      "epoch": 22.21,
+      "grad_norm": 1.5780655145645142,
+      "learning_rate": 8.888140161725068e-06,
+      "loss": 0.0047,
+      "step": 1510
+    },
+    {
+      "epoch": 22.35,
+      "grad_norm": 1.6800497770309448,
+      "learning_rate": 8.871293800539085e-06,
+      "loss": 0.0057,
+      "step": 1520
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 1.0676138401031494,
+      "learning_rate": 8.8544474393531e-06,
+      "loss": 0.0055,
+      "step": 1530
+    },
+    {
+      "epoch": 22.65,
+      "grad_norm": 2.0500409603118896,
+      "learning_rate": 8.837601078167117e-06,
+      "loss": 0.0059,
+      "step": 1540
+    },
+    {
+      "epoch": 22.79,
+      "grad_norm": 1.3901283740997314,
+      "learning_rate": 8.820754716981133e-06,
+      "loss": 0.0056,
+      "step": 1550
+    },
+    {
+      "epoch": 22.94,
+      "grad_norm": 1.7657601833343506,
+      "learning_rate": 8.803908355795148e-06,
+      "loss": 0.0069,
+      "step": 1560
+    },
+    {
+      "epoch": 23.09,
+      "grad_norm": 0.7901756763458252,
+      "learning_rate": 8.787061994609166e-06,
+      "loss": 0.0048,
+      "step": 1570
+    },
+    {
+      "epoch": 23.24,
+      "grad_norm": 1.3437069654464722,
+      "learning_rate": 8.770215633423182e-06,
+      "loss": 0.0052,
+      "step": 1580
+    },
+    {
+      "epoch": 23.38,
+      "grad_norm": 1.1439257860183716,
+      "learning_rate": 8.753369272237197e-06,
+      "loss": 0.0048,
+      "step": 1590
+    },
+    {
+      "epoch": 23.53,
+      "grad_norm": 2.013002872467041,
+      "learning_rate": 8.736522911051213e-06,
+      "loss": 0.005,
+      "step": 1600
+    },
+    {
+      "epoch": 23.68,
+      "grad_norm": 1.2465882301330566,
+      "learning_rate": 8.71967654986523e-06,
+      "loss": 0.0067,
+      "step": 1610
+    },
+    {
+      "epoch": 23.82,
+      "grad_norm": 2.2301642894744873,
+      "learning_rate": 8.702830188679245e-06,
+      "loss": 0.0058,
+      "step": 1620
+    },
+    {
+      "epoch": 23.97,
+      "grad_norm": 1.269247055053711,
+      "learning_rate": 8.685983827493262e-06,
+      "loss": 0.0055,
+      "step": 1630
+    },
+    {
+      "epoch": 24.12,
+      "grad_norm": 0.8886606693267822,
+      "learning_rate": 8.669137466307278e-06,
+      "loss": 0.0037,
+      "step": 1640
+    },
+    {
+      "epoch": 24.26,
+      "grad_norm": 1.7440249919891357,
+      "learning_rate": 8.652291105121294e-06,
+      "loss": 0.0049,
+      "step": 1650
+    },
+    {
+      "epoch": 24.41,
+      "grad_norm": 0.9428110122680664,
+      "learning_rate": 8.635444743935311e-06,
+      "loss": 0.0052,
+      "step": 1660
+    },
+    {
+      "epoch": 24.56,
+      "grad_norm": 1.336026906967163,
+      "learning_rate": 8.618598382749327e-06,
+      "loss": 0.0066,
+      "step": 1670
+    },
+    {
+      "epoch": 24.71,
+      "grad_norm": 0.7566521763801575,
+      "learning_rate": 8.601752021563343e-06,
+      "loss": 0.004,
+      "step": 1680
+    },
+    {
+      "epoch": 24.85,
+      "grad_norm": 1.2722612619400024,
+      "learning_rate": 8.58490566037736e-06,
+      "loss": 0.0039,
+      "step": 1690
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 0.7349913120269775,
+      "learning_rate": 8.568059299191376e-06,
+      "loss": 0.0048,
+      "step": 1700
+    },
+    {
+      "epoch": 25.15,
+      "grad_norm": 0.8350051641464233,
+      "learning_rate": 8.55121293800539e-06,
+      "loss": 0.004,
+      "step": 1710
+    },
+    {
+      "epoch": 25.29,
+      "grad_norm": 1.1201083660125732,
+      "learning_rate": 8.534366576819408e-06,
+      "loss": 0.0044,
+      "step": 1720
+    },
+    {
+      "epoch": 25.44,
+      "grad_norm": 0.9569421410560608,
+      "learning_rate": 8.517520215633423e-06,
+      "loss": 0.0048,
+      "step": 1730
+    },
+    {
+      "epoch": 25.59,
+      "grad_norm": 0.9667839407920837,
+      "learning_rate": 8.50067385444744e-06,
+      "loss": 0.0051,
+      "step": 1740
+    },
+    {
+      "epoch": 25.74,
+      "grad_norm": 0.6828237175941467,
+      "learning_rate": 8.483827493261457e-06,
+      "loss": 0.0061,
+      "step": 1750
+    },
+    {
+      "epoch": 25.88,
+      "grad_norm": 1.4523017406463623,
+      "learning_rate": 8.466981132075472e-06,
+      "loss": 0.0054,
+      "step": 1760
+    },
+    {
+      "epoch": 26.03,
+      "grad_norm": 1.0334619283676147,
+      "learning_rate": 8.450134770889488e-06,
+      "loss": 0.0034,
+      "step": 1770
+    },
+    {
+      "epoch": 26.18,
+      "grad_norm": 1.5895966291427612,
+      "learning_rate": 8.433288409703506e-06,
+      "loss": 0.0036,
+      "step": 1780
+    },
+    {
+      "epoch": 26.32,
+      "grad_norm": 1.0893361568450928,
+      "learning_rate": 8.416442048517522e-06,
+      "loss": 0.0041,
+      "step": 1790
+    },
+    {
+      "epoch": 26.47,
+      "grad_norm": 1.2191766500473022,
+      "learning_rate": 8.399595687331537e-06,
+      "loss": 0.0038,
+      "step": 1800
+    },
+    {
+      "epoch": 26.62,
+      "grad_norm": 1.8165545463562012,
+      "learning_rate": 8.382749326145553e-06,
+      "loss": 0.0038,
+      "step": 1810
+    },
+    {
+      "epoch": 26.76,
+      "grad_norm": 1.477317452430725,
+      "learning_rate": 8.365902964959569e-06,
+      "loss": 0.0044,
+      "step": 1820
+    },
+    {
+      "epoch": 26.91,
+      "grad_norm": 0.8306496739387512,
+      "learning_rate": 8.349056603773585e-06,
+      "loss": 0.0036,
+      "step": 1830
+    },
+    {
+      "epoch": 27.06,
+      "grad_norm": 0.9612523317337036,
+      "learning_rate": 8.332210242587602e-06,
+      "loss": 0.0031,
+      "step": 1840
+    },
+    {
+      "epoch": 27.21,
+      "grad_norm": 1.7324286699295044,
+      "learning_rate": 8.315363881401618e-06,
+      "loss": 0.0029,
+      "step": 1850
+    },
+    {
+      "epoch": 27.35,
+      "grad_norm": 1.0282984972000122,
+      "learning_rate": 8.298517520215634e-06,
+      "loss": 0.0038,
+      "step": 1860
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 0.7670619487762451,
+      "learning_rate": 8.281671159029651e-06,
+      "loss": 0.0044,
+      "step": 1870
+    },
+    {
+      "epoch": 27.65,
+      "grad_norm": 0.9554408192634583,
+      "learning_rate": 8.264824797843667e-06,
+      "loss": 0.0048,
+      "step": 1880
+    },
+    {
+      "epoch": 27.79,
+      "grad_norm": 1.2249259948730469,
+      "learning_rate": 8.247978436657683e-06,
+      "loss": 0.0038,
+      "step": 1890
+    },
+    {
+      "epoch": 27.94,
+      "grad_norm": 1.6148368120193481,
+      "learning_rate": 8.231132075471699e-06,
+      "loss": 0.0034,
+      "step": 1900
+    },
+    {
+      "epoch": 28.09,
+      "grad_norm": 1.3190560340881348,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.005,
+      "step": 1910
+    },
+    {
+      "epoch": 28.24,
+      "grad_norm": 1.7175958156585693,
+      "learning_rate": 8.197439353099732e-06,
+      "loss": 0.0037,
+      "step": 1920
+    },
+    {
+      "epoch": 28.38,
+      "grad_norm": 1.4584541320800781,
+      "learning_rate": 8.180592991913748e-06,
+      "loss": 0.0032,
+      "step": 1930
+    },
+    {
+      "epoch": 28.53,
+      "grad_norm": 1.2781018018722534,
+      "learning_rate": 8.163746630727763e-06,
+      "loss": 0.0038,
+      "step": 1940
+    },
+    {
+      "epoch": 28.68,
+      "grad_norm": 1.805674433708191,
+      "learning_rate": 8.146900269541779e-06,
+      "loss": 0.0044,
+      "step": 1950
+    },
+    {
+      "epoch": 28.82,
+      "grad_norm": 0.8535717129707336,
+      "learning_rate": 8.130053908355797e-06,
+      "loss": 0.0037,
+      "step": 1960
+    },
+    {
+      "epoch": 28.97,
+      "grad_norm": 1.6355313062667847,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 0.0033,
+      "step": 1970
+    },
+    {
+      "epoch": 29.12,
+      "grad_norm": 0.6453380584716797,
+      "learning_rate": 8.096361185983828e-06,
+      "loss": 0.0031,
+      "step": 1980
+    },
+    {
+      "epoch": 29.26,
+      "grad_norm": 1.806770920753479,
+      "learning_rate": 8.079514824797844e-06,
+      "loss": 0.0038,
+      "step": 1990
+    },
+    {
+      "epoch": 29.41,
+      "grad_norm": 0.49916043877601624,
+      "learning_rate": 8.06266846361186e-06,
+      "loss": 0.0032,
+      "step": 2000
+    },
+    {
+      "epoch": 29.41,
+      "eval_loss": 0.4711141288280487,
+      "eval_runtime": 394.2364,
+      "eval_samples_per_second": 1.913,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 42.29249011857708,
+      "step": 2000
+    },
+    {
+      "epoch": 29.56,
+      "grad_norm": 1.2711254358291626,
+      "learning_rate": 8.045822102425877e-06,
+      "loss": 0.0036,
+      "step": 2010
+    },
+    {
+      "epoch": 29.71,
+      "grad_norm": 1.2178030014038086,
+      "learning_rate": 8.028975741239893e-06,
+      "loss": 0.0042,
+      "step": 2020
+    },
+    {
+      "epoch": 29.85,
+      "grad_norm": 1.2590820789337158,
+      "learning_rate": 8.012129380053909e-06,
+      "loss": 0.0033,
+      "step": 2030
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.7739225625991821,
+      "learning_rate": 7.995283018867925e-06,
+      "loss": 0.0053,
+      "step": 2040
+    },
+    {
+      "epoch": 30.15,
+      "grad_norm": 1.882087230682373,
+      "learning_rate": 7.978436657681942e-06,
+      "loss": 0.0043,
+      "step": 2050
+    },
+    {
+      "epoch": 30.29,
+      "grad_norm": 1.1462137699127197,
+      "learning_rate": 7.961590296495958e-06,
+      "loss": 0.0027,
+      "step": 2060
+    },
+    {
+      "epoch": 30.44,
+      "grad_norm": 1.1864248514175415,
+      "learning_rate": 7.944743935309974e-06,
+      "loss": 0.003,
+      "step": 2070
+    },
+    {
+      "epoch": 30.59,
+      "grad_norm": 0.4730267822742462,
+      "learning_rate": 7.92789757412399e-06,
+      "loss": 0.0026,
+      "step": 2080
+    },
+    {
+      "epoch": 30.74,
+      "grad_norm": 0.8079692125320435,
+      "learning_rate": 7.911051212938005e-06,
+      "loss": 0.002,
+      "step": 2090
+    },
+    {
+      "epoch": 30.88,
+      "grad_norm": 0.9878515005111694,
+      "learning_rate": 7.894204851752023e-06,
+      "loss": 0.0039,
+      "step": 2100
+    },
+    {
+      "epoch": 31.03,
+      "grad_norm": 0.26896196603775024,
+      "learning_rate": 7.877358490566038e-06,
+      "loss": 0.0033,
+      "step": 2110
+    },
+    {
+      "epoch": 31.18,
+      "grad_norm": 0.5570561289787292,
+      "learning_rate": 7.860512129380054e-06,
+      "loss": 0.0041,
+      "step": 2120
+    },
+    {
+      "epoch": 31.32,
+      "grad_norm": 1.9286779165267944,
+      "learning_rate": 7.843665768194072e-06,
+      "loss": 0.0031,
+      "step": 2130
+    },
+    {
+      "epoch": 31.47,
+      "grad_norm": 0.5159671306610107,
+      "learning_rate": 7.826819407008087e-06,
+      "loss": 0.002,
+      "step": 2140
+    },
+    {
+      "epoch": 31.62,
+      "grad_norm": 0.811177670955658,
+      "learning_rate": 7.809973045822103e-06,
+      "loss": 0.003,
+      "step": 2150
+    },
+    {
+      "epoch": 31.76,
+      "grad_norm": 0.8233473300933838,
+      "learning_rate": 7.793126684636119e-06,
+      "loss": 0.0023,
+      "step": 2160
+    },
+    {
+      "epoch": 31.91,
+      "grad_norm": 0.3896019756793976,
+      "learning_rate": 7.776280323450135e-06,
+      "loss": 0.0021,
+      "step": 2170
+    },
+    {
+      "epoch": 32.06,
+      "grad_norm": 0.37738358974456787,
+      "learning_rate": 7.75943396226415e-06,
+      "loss": 0.0024,
+      "step": 2180
+    },
+    {
+      "epoch": 32.21,
+      "grad_norm": 1.1048047542572021,
+      "learning_rate": 7.742587601078168e-06,
+      "loss": 0.003,
+      "step": 2190
+    },
+    {
+      "epoch": 32.35,
+      "grad_norm": 0.4888085126876831,
+      "learning_rate": 7.725741239892184e-06,
+      "loss": 0.0018,
+      "step": 2200
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 0.8060516119003296,
+      "learning_rate": 7.7088948787062e-06,
+      "loss": 0.0019,
+      "step": 2210
+    },
+    {
+      "epoch": 32.65,
+      "grad_norm": 0.8470051884651184,
+      "learning_rate": 7.692048517520217e-06,
+      "loss": 0.0019,
+      "step": 2220
+    },
+    {
+      "epoch": 32.79,
+      "grad_norm": 0.9551669955253601,
+      "learning_rate": 7.675202156334233e-06,
+      "loss": 0.0027,
+      "step": 2230
+    },
+    {
+      "epoch": 32.94,
+      "grad_norm": 1.0600543022155762,
+      "learning_rate": 7.658355795148249e-06,
+      "loss": 0.0025,
+      "step": 2240
+    },
+    {
+      "epoch": 33.09,
+      "grad_norm": 0.4787401258945465,
+      "learning_rate": 7.641509433962266e-06,
+      "loss": 0.0027,
+      "step": 2250
+    },
+    {
+      "epoch": 33.24,
+      "grad_norm": 0.7312725782394409,
+      "learning_rate": 7.62466307277628e-06,
+      "loss": 0.0024,
+      "step": 2260
+    },
+    {
+      "epoch": 33.38,
+      "grad_norm": 1.3341702222824097,
+      "learning_rate": 7.607816711590297e-06,
+      "loss": 0.0022,
+      "step": 2270
+    },
+    {
+      "epoch": 33.53,
+      "grad_norm": 0.6375266909599304,
+      "learning_rate": 7.5909703504043134e-06,
+      "loss": 0.0029,
+      "step": 2280
+    },
+    {
+      "epoch": 33.68,
+      "grad_norm": 0.8027037978172302,
+      "learning_rate": 7.574123989218329e-06,
+      "loss": 0.0019,
+      "step": 2290
+    },
+    {
+      "epoch": 33.82,
+      "grad_norm": 0.45866358280181885,
+      "learning_rate": 7.557277628032346e-06,
+      "loss": 0.002,
+      "step": 2300
+    },
+    {
+      "epoch": 33.97,
+      "grad_norm": 0.4257550835609436,
+      "learning_rate": 7.540431266846362e-06,
+      "loss": 0.0022,
+      "step": 2310
+    },
+    {
+      "epoch": 34.12,
+      "grad_norm": 0.6710132360458374,
+      "learning_rate": 7.523584905660378e-06,
+      "loss": 0.0027,
+      "step": 2320
+    },
+    {
+      "epoch": 34.26,
+      "grad_norm": 0.7800111770629883,
+      "learning_rate": 7.506738544474395e-06,
+      "loss": 0.0022,
+      "step": 2330
+    },
+    {
+      "epoch": 34.41,
+      "grad_norm": 0.5604603290557861,
+      "learning_rate": 7.489892183288411e-06,
+      "loss": 0.0028,
+      "step": 2340
+    },
+    {
+      "epoch": 34.56,
+      "grad_norm": 1.3979426622390747,
+      "learning_rate": 7.473045822102426e-06,
+      "loss": 0.0024,
+      "step": 2350
+    },
+    {
+      "epoch": 34.71,
+      "grad_norm": 1.4295878410339355,
+      "learning_rate": 7.456199460916442e-06,
+      "loss": 0.0025,
+      "step": 2360
+    },
+    {
+      "epoch": 34.85,
+      "grad_norm": 0.8446325659751892,
+      "learning_rate": 7.439353099730459e-06,
+      "loss": 0.0039,
+      "step": 2370
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 0.7062538266181946,
+      "learning_rate": 7.422506738544475e-06,
+      "loss": 0.0024,
+      "step": 2380
+    },
+    {
+      "epoch": 35.15,
+      "grad_norm": 0.42034047842025757,
+      "learning_rate": 7.405660377358491e-06,
+      "loss": 0.0015,
+      "step": 2390
+    },
+    {
+      "epoch": 35.29,
+      "grad_norm": 0.21419081091880798,
+      "learning_rate": 7.388814016172508e-06,
+      "loss": 0.0015,
+      "step": 2400
+    },
+    {
+      "epoch": 35.44,
+      "grad_norm": 0.2849736213684082,
+      "learning_rate": 7.371967654986524e-06,
+      "loss": 0.0008,
+      "step": 2410
+    },
+    {
+      "epoch": 35.59,
+      "grad_norm": 0.4012085795402527,
+      "learning_rate": 7.35512129380054e-06,
+      "loss": 0.0015,
+      "step": 2420
+    },
+    {
+      "epoch": 35.74,
+      "grad_norm": 0.12295886129140854,
+      "learning_rate": 7.338274932614556e-06,
+      "loss": 0.0016,
+      "step": 2430
+    },
+    {
+      "epoch": 35.88,
+      "grad_norm": 0.5100615620613098,
+      "learning_rate": 7.321428571428572e-06,
+      "loss": 0.0017,
+      "step": 2440
+    },
+    {
+      "epoch": 36.03,
+      "grad_norm": 0.19803810119628906,
+      "learning_rate": 7.304582210242588e-06,
+      "loss": 0.0013,
+      "step": 2450
+    },
+    {
+      "epoch": 36.18,
+      "grad_norm": 0.45984306931495667,
+      "learning_rate": 7.287735849056604e-06,
+      "loss": 0.0007,
+      "step": 2460
+    },
+    {
+      "epoch": 36.32,
+      "grad_norm": 1.1022437810897827,
+      "learning_rate": 7.27088948787062e-06,
+      "loss": 0.0012,
+      "step": 2470
+    },
+    {
+      "epoch": 36.47,
+      "grad_norm": 0.20512618124485016,
+      "learning_rate": 7.254043126684637e-06,
+      "loss": 0.0017,
+      "step": 2480
+    },
+    {
+      "epoch": 36.62,
+      "grad_norm": 0.11672288924455643,
+      "learning_rate": 7.237196765498653e-06,
+      "loss": 0.0009,
+      "step": 2490
+    },
+    {
+      "epoch": 36.76,
+      "grad_norm": 0.41004320979118347,
+      "learning_rate": 7.220350404312669e-06,
+      "loss": 0.0005,
+      "step": 2500
+    },
+    {
+      "epoch": 36.76,
+      "eval_loss": 0.48888495564460754,
+      "eval_runtime": 395.4163,
+      "eval_samples_per_second": 1.907,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 41.22529644268775,
+      "step": 2500
+    },
+    {
+      "epoch": 36.91,
+      "grad_norm": 0.41586098074913025,
+      "learning_rate": 7.203504043126686e-06,
+      "loss": 0.0006,
+      "step": 2510
+    },
+    {
+      "epoch": 37.06,
+      "grad_norm": 0.05540100112557411,
+      "learning_rate": 7.1866576819407015e-06,
+      "loss": 0.0004,
+      "step": 2520
+    },
+    {
+      "epoch": 37.21,
+      "grad_norm": 0.20431630313396454,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 0.0005,
+      "step": 2530
+    },
+    {
+      "epoch": 37.35,
+      "grad_norm": 0.13198405504226685,
+      "learning_rate": 7.152964959568733e-06,
+      "loss": 0.0005,
+      "step": 2540
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 1.1269527673721313,
+      "learning_rate": 7.13611859838275e-06,
+      "loss": 0.0006,
+      "step": 2550
+    },
+    {
+      "epoch": 37.65,
+      "grad_norm": 0.06169961765408516,
+      "learning_rate": 7.119272237196766e-06,
+      "loss": 0.0003,
+      "step": 2560
+    },
+    {
+      "epoch": 37.79,
+      "grad_norm": 1.2373998165130615,
+      "learning_rate": 7.102425876010782e-06,
+      "loss": 0.0007,
+      "step": 2570
+    },
+    {
+      "epoch": 37.94,
+      "grad_norm": 0.730436384677887,
+      "learning_rate": 7.085579514824799e-06,
+      "loss": 0.0006,
+      "step": 2580
+    },
+    {
+      "epoch": 38.09,
+      "grad_norm": 0.2517056465148926,
+      "learning_rate": 7.0687331536388145e-06,
+      "loss": 0.0009,
+      "step": 2590
+    },
+    {
+      "epoch": 38.24,
+      "grad_norm": 0.24311576783657074,
+      "learning_rate": 7.051886792452831e-06,
+      "loss": 0.0008,
+      "step": 2600
+    },
+    {
+      "epoch": 38.38,
+      "grad_norm": 0.46015554666519165,
+      "learning_rate": 7.035040431266848e-06,
+      "loss": 0.0008,
+      "step": 2610
+    },
+    {
+      "epoch": 38.53,
+      "grad_norm": 0.10832543671131134,
+      "learning_rate": 7.018194070080863e-06,
+      "loss": 0.0011,
+      "step": 2620
+    },
+    {
+      "epoch": 38.68,
+      "grad_norm": 0.06158292293548584,
+      "learning_rate": 7.0013477088948785e-06,
+      "loss": 0.0011,
+      "step": 2630
+    },
+    {
+      "epoch": 38.82,
+      "grad_norm": 0.8122970461845398,
+      "learning_rate": 6.984501347708895e-06,
+      "loss": 0.0009,
+      "step": 2640
+    },
+    {
+      "epoch": 38.97,
+      "grad_norm": 0.7663437128067017,
+      "learning_rate": 6.967654986522912e-06,
+      "loss": 0.0008,
+      "step": 2650
+    },
+    {
+      "epoch": 39.12,
+      "grad_norm": 0.145246684551239,
+      "learning_rate": 6.9508086253369275e-06,
+      "loss": 0.0007,
+      "step": 2660
+    },
+    {
+      "epoch": 39.26,
+      "grad_norm": 0.5487465858459473,
+      "learning_rate": 6.933962264150944e-06,
+      "loss": 0.0016,
+      "step": 2670
+    },
+    {
+      "epoch": 39.41,
+      "grad_norm": 0.9387257695198059,
+      "learning_rate": 6.917115902964961e-06,
+      "loss": 0.0009,
+      "step": 2680
+    },
+    {
+      "epoch": 39.56,
+      "grad_norm": 0.31192561984062195,
+      "learning_rate": 6.9002695417789766e-06,
+      "loss": 0.0008,
+      "step": 2690
+    },
+    {
+      "epoch": 39.71,
+      "grad_norm": 0.0915888324379921,
+      "learning_rate": 6.883423180592993e-06,
+      "loss": 0.0005,
+      "step": 2700
+    },
+    {
+      "epoch": 39.85,
+      "grad_norm": 0.10339212417602539,
+      "learning_rate": 6.866576819407008e-06,
+      "loss": 0.0005,
+      "step": 2710
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.07240493595600128,
+      "learning_rate": 6.849730458221025e-06,
+      "loss": 0.0008,
+      "step": 2720
+    },
+    {
+      "epoch": 40.15,
+      "grad_norm": 0.08370707184076309,
+      "learning_rate": 6.8328840970350405e-06,
+      "loss": 0.0009,
+      "step": 2730
+    },
+    {
+      "epoch": 40.29,
+      "grad_norm": 0.07619204372167587,
+      "learning_rate": 6.816037735849057e-06,
+      "loss": 0.0005,
+      "step": 2740
+    },
+    {
+      "epoch": 40.44,
+      "grad_norm": 0.4352577030658722,
+      "learning_rate": 6.799191374663073e-06,
+      "loss": 0.0016,
+      "step": 2750
+    },
+    {
+      "epoch": 40.59,
+      "grad_norm": 0.6862916946411133,
+      "learning_rate": 6.78234501347709e-06,
+      "loss": 0.0006,
+      "step": 2760
+    },
+    {
+      "epoch": 40.74,
+      "grad_norm": 0.2726142406463623,
+      "learning_rate": 6.765498652291106e-06,
+      "loss": 0.0008,
+      "step": 2770
+    },
+    {
+      "epoch": 40.88,
+      "grad_norm": 0.1969071477651596,
+      "learning_rate": 6.748652291105122e-06,
+      "loss": 0.0005,
+      "step": 2780
+    },
+    {
+      "epoch": 41.03,
+      "grad_norm": 0.341925710439682,
+      "learning_rate": 6.731805929919139e-06,
+      "loss": 0.0007,
+      "step": 2790
+    },
+    {
+      "epoch": 41.18,
+      "grad_norm": 1.3179571628570557,
+      "learning_rate": 6.7149595687331536e-06,
+      "loss": 0.0005,
+      "step": 2800
+    },
+    {
+      "epoch": 41.32,
+      "grad_norm": 0.5500785112380981,
+      "learning_rate": 6.69811320754717e-06,
+      "loss": 0.0014,
+      "step": 2810
+    },
+    {
+      "epoch": 41.47,
+      "grad_norm": 0.30600807070732117,
+      "learning_rate": 6.681266846361186e-06,
+      "loss": 0.0008,
+      "step": 2820
+    },
+    {
+      "epoch": 41.62,
+      "grad_norm": 0.2631785571575165,
+      "learning_rate": 6.664420485175203e-06,
+      "loss": 0.001,
+      "step": 2830
+    },
+    {
+      "epoch": 41.76,
+      "grad_norm": 0.8868472576141357,
+      "learning_rate": 6.647574123989219e-06,
+      "loss": 0.0009,
+      "step": 2840
+    },
+    {
+      "epoch": 41.91,
+      "grad_norm": 1.278052806854248,
+      "learning_rate": 6.630727762803235e-06,
+      "loss": 0.0011,
+      "step": 2850
+    },
+    {
+      "epoch": 42.06,
+      "grad_norm": 0.1165945902466774,
+      "learning_rate": 6.613881401617252e-06,
+      "loss": 0.0014,
+      "step": 2860
+    },
+    {
+      "epoch": 42.21,
+      "grad_norm": 0.37418603897094727,
+      "learning_rate": 6.597035040431267e-06,
+      "loss": 0.0009,
+      "step": 2870
+    },
+    {
+      "epoch": 42.35,
+      "grad_norm": 2.285724401473999,
+      "learning_rate": 6.580188679245284e-06,
+      "loss": 0.0013,
+      "step": 2880
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 0.48481008410453796,
+      "learning_rate": 6.563342318059299e-06,
+      "loss": 0.0012,
+      "step": 2890
+    },
+    {
+      "epoch": 42.65,
+      "grad_norm": 0.330135703086853,
+      "learning_rate": 6.546495956873316e-06,
+      "loss": 0.0007,
+      "step": 2900
+    },
+    {
+      "epoch": 42.79,
+      "grad_norm": 0.5341326594352722,
+      "learning_rate": 6.529649595687332e-06,
+      "loss": 0.0008,
+      "step": 2910
+    },
+    {
+      "epoch": 42.94,
+      "grad_norm": 1.2509163618087769,
+      "learning_rate": 6.512803234501348e-06,
+      "loss": 0.0016,
+      "step": 2920
+    },
+    {
+      "epoch": 43.09,
+      "grad_norm": 1.3307183980941772,
+      "learning_rate": 6.495956873315365e-06,
+      "loss": 0.0015,
+      "step": 2930
+    },
+    {
+      "epoch": 43.24,
+      "grad_norm": 0.4615732729434967,
+      "learning_rate": 6.47911051212938e-06,
+      "loss": 0.0011,
+      "step": 2940
+    },
+    {
+      "epoch": 43.38,
+      "grad_norm": 0.3208690583705902,
+      "learning_rate": 6.462264150943397e-06,
+      "loss": 0.001,
+      "step": 2950
+    },
+    {
+      "epoch": 43.53,
+      "grad_norm": 0.6121777892112732,
+      "learning_rate": 6.445417789757414e-06,
+      "loss": 0.0012,
+      "step": 2960
+    },
+    {
+      "epoch": 43.68,
+      "grad_norm": 0.8947989344596863,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.0021,
+      "step": 2970
+    },
+    {
+      "epoch": 43.82,
+      "grad_norm": 1.4063951969146729,
+      "learning_rate": 6.411725067385444e-06,
+      "loss": 0.0022,
+      "step": 2980
+    },
+    {
+      "epoch": 43.97,
+      "grad_norm": 0.9614599347114563,
+      "learning_rate": 6.394878706199461e-06,
+      "loss": 0.0022,
+      "step": 2990
+    },
+    {
+      "epoch": 44.12,
+      "grad_norm": 0.3544367551803589,
+      "learning_rate": 6.378032345013478e-06,
+      "loss": 0.0012,
+      "step": 3000
+    },
+    {
+      "epoch": 44.12,
+      "eval_loss": 0.4919167459011078,
+      "eval_runtime": 412.7274,
+      "eval_samples_per_second": 1.827,
+      "eval_steps_per_second": 0.019,
+      "eval_wer": 42.03557312252965,
+      "step": 3000
+    },
+    {
+      "epoch": 44.26,
+      "grad_norm": 0.3670799732208252,
+      "learning_rate": 6.3611859838274934e-06,
+      "loss": 0.0011,
+      "step": 3010
+    },
+    {
+      "epoch": 44.41,
+      "grad_norm": 0.8639935255050659,
+      "learning_rate": 6.34433962264151e-06,
+      "loss": 0.0021,
+      "step": 3020
+    },
+    {
+      "epoch": 44.56,
+      "grad_norm": 0.8535746932029724,
+      "learning_rate": 6.327493261455526e-06,
+      "loss": 0.0023,
+      "step": 3030
+    },
+    {
+      "epoch": 44.71,
+      "grad_norm": 0.791280210018158,
+      "learning_rate": 6.3106469002695425e-06,
+      "loss": 0.0017,
+      "step": 3040
+    },
+    {
+      "epoch": 44.85,
+      "grad_norm": 1.2309428453445435,
+      "learning_rate": 6.293800539083559e-06,
+      "loss": 0.0022,
+      "step": 3050
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 2.02470326423645,
+      "learning_rate": 6.276954177897575e-06,
+      "loss": 0.0023,
+      "step": 3060
+    },
+    {
+      "epoch": 45.15,
+      "grad_norm": 0.6730286478996277,
+      "learning_rate": 6.260107816711591e-06,
+      "loss": 0.0014,
+      "step": 3070
+    },
+    {
+      "epoch": 45.29,
+      "grad_norm": 0.6371138095855713,
+      "learning_rate": 6.2432614555256064e-06,
+      "loss": 0.0013,
+      "step": 3080
+    },
+    {
+      "epoch": 45.44,
+      "grad_norm": 0.8862447738647461,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 0.002,
+      "step": 3090
+    },
+    {
+      "epoch": 45.59,
+      "grad_norm": 0.600082516670227,
+      "learning_rate": 6.209568733153639e-06,
+      "loss": 0.0015,
+      "step": 3100
+    },
+    {
+      "epoch": 45.74,
+      "grad_norm": 1.0069173574447632,
+      "learning_rate": 6.1927223719676555e-06,
+      "loss": 0.0011,
+      "step": 3110
+    },
+    {
+      "epoch": 45.88,
+      "grad_norm": 0.8306257724761963,
+      "learning_rate": 6.175876010781672e-06,
+      "loss": 0.0017,
+      "step": 3120
+    },
+    {
+      "epoch": 46.03,
+      "grad_norm": 0.0885746031999588,
+      "learning_rate": 6.159029649595688e-06,
+      "loss": 0.002,
+      "step": 3130
+    },
+    {
+      "epoch": 46.18,
+      "grad_norm": 0.2680250108242035,
+      "learning_rate": 6.1421832884097045e-06,
+      "loss": 0.0011,
+      "step": 3140
+    },
+    {
+      "epoch": 46.32,
+      "grad_norm": 0.5739225745201111,
+      "learning_rate": 6.12533692722372e-06,
+      "loss": 0.0006,
+      "step": 3150
+    },
+    {
+      "epoch": 46.47,
+      "grad_norm": 0.09838727116584778,
+      "learning_rate": 6.108490566037736e-06,
+      "loss": 0.001,
+      "step": 3160
+    },
+    {
+      "epoch": 46.62,
+      "grad_norm": 0.6324687600135803,
+      "learning_rate": 6.091644204851752e-06,
+      "loss": 0.0012,
+      "step": 3170
+    },
+    {
+      "epoch": 46.76,
+      "grad_norm": 0.8236867189407349,
+      "learning_rate": 6.0747978436657685e-06,
+      "loss": 0.0018,
+      "step": 3180
+    },
+    {
+      "epoch": 46.91,
+      "grad_norm": 1.935333251953125,
+      "learning_rate": 6.057951482479785e-06,
+      "loss": 0.0028,
+      "step": 3190
+    },
+    {
+      "epoch": 47.06,
+      "grad_norm": 0.9879902601242065,
+      "learning_rate": 6.041105121293801e-06,
+      "loss": 0.0025,
+      "step": 3200
+    },
+    {
+      "epoch": 47.21,
+      "grad_norm": 0.7782288193702698,
+      "learning_rate": 6.0242587601078175e-06,
+      "loss": 0.0017,
+      "step": 3210
+    },
+    {
+      "epoch": 47.35,
+      "grad_norm": 0.9317180514335632,
+      "learning_rate": 6.007412398921833e-06,
+      "loss": 0.0014,
+      "step": 3220
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 0.4948224127292633,
+      "learning_rate": 5.99056603773585e-06,
+      "loss": 0.0012,
+      "step": 3230
+    },
+    {
+      "epoch": 47.65,
+      "grad_norm": 0.5893387198448181,
+      "learning_rate": 5.9737196765498666e-06,
+      "loss": 0.0012,
+      "step": 3240
+    },
+    {
+      "epoch": 47.79,
+      "grad_norm": 0.22002609074115753,
+      "learning_rate": 5.9568733153638815e-06,
+      "loss": 0.0015,
+      "step": 3250
+    },
+    {
+      "epoch": 47.94,
+      "grad_norm": 0.4735713005065918,
+      "learning_rate": 5.940026954177897e-06,
+      "loss": 0.0008,
+      "step": 3260
+    },
+    {
+      "epoch": 48.09,
+      "grad_norm": 0.8969672918319702,
+      "learning_rate": 5.923180592991914e-06,
+      "loss": 0.0014,
+      "step": 3270
+    },
+    {
+      "epoch": 48.24,
+      "grad_norm": 0.6368213295936584,
+      "learning_rate": 5.9063342318059305e-06,
+      "loss": 0.0014,
+      "step": 3280
+    },
+    {
+      "epoch": 48.38,
+      "grad_norm": 0.1679704338312149,
+      "learning_rate": 5.889487870619946e-06,
+      "loss": 0.0013,
+      "step": 3290
+    },
+    {
+      "epoch": 48.53,
+      "grad_norm": 0.9470486044883728,
+      "learning_rate": 5.872641509433963e-06,
+      "loss": 0.0017,
+      "step": 3300
+    },
+    {
+      "epoch": 48.68,
+      "grad_norm": 0.08789900690317154,
+      "learning_rate": 5.855795148247979e-06,
+      "loss": 0.0007,
+      "step": 3310
+    },
+    {
+      "epoch": 48.82,
+      "grad_norm": 0.24826857447624207,
+      "learning_rate": 5.838948787061995e-06,
+      "loss": 0.0009,
+      "step": 3320
+    },
+    {
+      "epoch": 48.97,
+      "grad_norm": 0.11028200387954712,
+      "learning_rate": 5.822102425876012e-06,
+      "loss": 0.0009,
+      "step": 3330
+    },
+    {
+      "epoch": 49.12,
+      "grad_norm": 0.11782591044902802,
+      "learning_rate": 5.805256064690027e-06,
+      "loss": 0.0008,
+      "step": 3340
+    },
+    {
+      "epoch": 49.26,
+      "grad_norm": 0.3089038133621216,
+      "learning_rate": 5.7884097035040435e-06,
+      "loss": 0.0007,
+      "step": 3350
+    },
+    {
+      "epoch": 49.41,
+      "grad_norm": 0.5864781737327576,
+      "learning_rate": 5.771563342318059e-06,
+      "loss": 0.0006,
+      "step": 3360
+    },
+    {
+      "epoch": 49.56,
+      "grad_norm": 0.6767354607582092,
+      "learning_rate": 5.754716981132076e-06,
+      "loss": 0.0004,
+      "step": 3370
+    },
+    {
+      "epoch": 49.71,
+      "grad_norm": 0.7366251349449158,
+      "learning_rate": 5.737870619946092e-06,
+      "loss": 0.0006,
+      "step": 3380
+    },
+    {
+      "epoch": 49.85,
+      "grad_norm": 0.460439532995224,
+      "learning_rate": 5.721024258760108e-06,
+      "loss": 0.0006,
+      "step": 3390
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 0.2541469633579254,
+      "learning_rate": 5.704177897574125e-06,
+      "loss": 0.0011,
+      "step": 3400
+    },
+    {
+      "epoch": 50.15,
+      "grad_norm": 0.25263655185699463,
+      "learning_rate": 5.687331536388141e-06,
+      "loss": 0.0009,
+      "step": 3410
+    },
+    {
+      "epoch": 50.29,
+      "grad_norm": 0.3167230188846588,
+      "learning_rate": 5.670485175202157e-06,
+      "loss": 0.0005,
+      "step": 3420
+    },
+    {
+      "epoch": 50.44,
+      "grad_norm": 0.051995839923620224,
+      "learning_rate": 5.653638814016172e-06,
+      "loss": 0.0005,
+      "step": 3430
+    },
+    {
+      "epoch": 50.59,
+      "grad_norm": 0.23166793584823608,
+      "learning_rate": 5.636792452830189e-06,
+      "loss": 0.0012,
+      "step": 3440
+    },
+    {
+      "epoch": 50.74,
+      "grad_norm": 0.46085381507873535,
+      "learning_rate": 5.619946091644205e-06,
+      "loss": 0.0011,
+      "step": 3450
+    },
+    {
+      "epoch": 50.88,
+      "grad_norm": 0.07778172194957733,
+      "learning_rate": 5.603099730458221e-06,
+      "loss": 0.0006,
+      "step": 3460
+    },
+    {
+      "epoch": 51.03,
+      "grad_norm": 0.23257611691951752,
+      "learning_rate": 5.586253369272238e-06,
+      "loss": 0.0009,
+      "step": 3470
+    },
+    {
+      "epoch": 51.18,
+      "grad_norm": 0.6834172606468201,
+      "learning_rate": 5.569407008086254e-06,
+      "loss": 0.0006,
+      "step": 3480
+    },
+    {
+      "epoch": 51.32,
+      "grad_norm": 0.028312204405665398,
+      "learning_rate": 5.55256064690027e-06,
+      "loss": 0.0008,
+      "step": 3490
+    },
+    {
+      "epoch": 51.47,
+      "grad_norm": 0.053247835487127304,
+      "learning_rate": 5.535714285714286e-06,
+      "loss": 0.0004,
+      "step": 3500
+    },
+    {
+      "epoch": 51.47,
+      "eval_loss": 0.5013700127601624,
+      "eval_runtime": 393.4347,
+      "eval_samples_per_second": 1.916,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 40.39525691699605,
+      "step": 3500
+    },
+    {
+      "epoch": 51.62,
+      "grad_norm": 0.09007065743207932,
+      "learning_rate": 5.518867924528303e-06,
+      "loss": 0.0003,
+      "step": 3510
+    },
+    {
+      "epoch": 51.76,
+      "grad_norm": 0.03709765151143074,
+      "learning_rate": 5.502021563342318e-06,
+      "loss": 0.0003,
+      "step": 3520
+    },
+    {
+      "epoch": 51.91,
+      "grad_norm": 0.024563109502196312,
+      "learning_rate": 5.485175202156334e-06,
+      "loss": 0.0002,
+      "step": 3530
+    },
+    {
+      "epoch": 52.06,
+      "grad_norm": 0.38505399227142334,
+      "learning_rate": 5.46832884097035e-06,
+      "loss": 0.0007,
+      "step": 3540
+    },
+    {
+      "epoch": 52.21,
+      "grad_norm": 0.041454803198575974,
+      "learning_rate": 5.451482479784367e-06,
+      "loss": 0.0002,
+      "step": 3550
+    },
+    {
+      "epoch": 52.35,
+      "grad_norm": 0.03118882142007351,
+      "learning_rate": 5.4346361185983834e-06,
+      "loss": 0.0002,
+      "step": 3560
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 0.1050623282790184,
+      "learning_rate": 5.417789757412399e-06,
+      "loss": 0.0006,
+      "step": 3570
+    },
+    {
+      "epoch": 52.65,
+      "grad_norm": 0.019064374268054962,
+      "learning_rate": 5.400943396226416e-06,
+      "loss": 0.0001,
+      "step": 3580
+    },
+    {
+      "epoch": 52.79,
+      "grad_norm": 0.020824899896979332,
+      "learning_rate": 5.384097035040432e-06,
+      "loss": 0.0001,
+      "step": 3590
+    },
+    {
+      "epoch": 52.94,
+      "grad_norm": 0.019172310829162598,
+      "learning_rate": 5.367250673854448e-06,
+      "loss": 0.0002,
+      "step": 3600
+    },
+    {
+      "epoch": 53.09,
+      "grad_norm": 0.015175268054008484,
+      "learning_rate": 5.350404312668463e-06,
+      "loss": 0.0003,
+      "step": 3610
+    },
+    {
+      "epoch": 53.24,
+      "grad_norm": 0.03325885161757469,
+      "learning_rate": 5.33355795148248e-06,
+      "loss": 0.0004,
+      "step": 3620
+    },
+    {
+      "epoch": 53.38,
+      "grad_norm": 0.02436959184706211,
+      "learning_rate": 5.3167115902964964e-06,
+      "loss": 0.0002,
+      "step": 3630
+    },
+    {
+      "epoch": 53.53,
+      "grad_norm": 0.017924895510077477,
+      "learning_rate": 5.299865229110512e-06,
+      "loss": 0.0001,
+      "step": 3640
+    },
+    {
+      "epoch": 53.68,
+      "grad_norm": 0.02767987549304962,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 0.0001,
+      "step": 3650
+    },
+    {
+      "epoch": 53.82,
+      "grad_norm": 0.01485258899629116,
+      "learning_rate": 5.266172506738545e-06,
+      "loss": 0.0003,
+      "step": 3660
+    },
+    {
+      "epoch": 53.97,
+      "grad_norm": 0.013557419180870056,
+      "learning_rate": 5.249326145552561e-06,
+      "loss": 0.0001,
+      "step": 3670
+    },
+    {
+      "epoch": 54.12,
+      "grad_norm": 0.010153830982744694,
+      "learning_rate": 5.232479784366578e-06,
+      "loss": 0.0002,
+      "step": 3680
+    },
+    {
+      "epoch": 54.26,
+      "grad_norm": 0.01170294638723135,
+      "learning_rate": 5.215633423180594e-06,
+      "loss": 0.0001,
+      "step": 3690
+    },
+    {
+      "epoch": 54.41,
+      "grad_norm": 0.008615361526608467,
+      "learning_rate": 5.19878706199461e-06,
+      "loss": 0.0001,
+      "step": 3700
+    },
+    {
+      "epoch": 54.56,
+      "grad_norm": 0.00960615649819374,
+      "learning_rate": 5.181940700808625e-06,
+      "loss": 0.0001,
+      "step": 3710
+    },
+    {
+      "epoch": 54.71,
+      "grad_norm": 0.008886042051017284,
+      "learning_rate": 5.165094339622642e-06,
+      "loss": 0.0001,
+      "step": 3720
+    },
+    {
+      "epoch": 54.85,
+      "grad_norm": 0.014573791064321995,
+      "learning_rate": 5.148247978436658e-06,
+      "loss": 0.0001,
+      "step": 3730
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 0.008440797217190266,
+      "learning_rate": 5.131401617250674e-06,
+      "loss": 0.0001,
+      "step": 3740
+    },
+    {
+      "epoch": 55.15,
+      "grad_norm": 0.010030324570834637,
+      "learning_rate": 5.114555256064691e-06,
+      "loss": 0.0001,
+      "step": 3750
+    },
+    {
+      "epoch": 55.29,
+      "grad_norm": 0.0625586286187172,
+      "learning_rate": 5.097708894878707e-06,
+      "loss": 0.0001,
+      "step": 3760
+    },
+    {
+      "epoch": 55.44,
+      "grad_norm": 0.011248340830206871,
+      "learning_rate": 5.080862533692723e-06,
+      "loss": 0.0001,
+      "step": 3770
+    },
+    {
+      "epoch": 55.59,
+      "grad_norm": 0.006955728866159916,
+      "learning_rate": 5.064016172506739e-06,
+      "loss": 0.0001,
+      "step": 3780
+    },
+    {
+      "epoch": 55.74,
+      "grad_norm": 0.00837327353656292,
+      "learning_rate": 5.047169811320756e-06,
+      "loss": 0.0001,
+      "step": 3790
+    },
+    {
+      "epoch": 55.88,
+      "grad_norm": 0.008046143688261509,
+      "learning_rate": 5.030323450134771e-06,
+      "loss": 0.0001,
+      "step": 3800
+    },
+    {
+      "epoch": 56.03,
+      "grad_norm": 0.007838546298444271,
+      "learning_rate": 5.013477088948787e-06,
+      "loss": 0.0001,
+      "step": 3810
+    },
+    {
+      "epoch": 56.18,
+      "grad_norm": 0.006525520700961351,
+      "learning_rate": 4.996630727762803e-06,
+      "loss": 0.0001,
+      "step": 3820
+    },
+    {
+      "epoch": 56.32,
+      "grad_norm": 0.005834328010678291,
+      "learning_rate": 4.97978436657682e-06,
+      "loss": 0.0001,
+      "step": 3830
+    },
+    {
+      "epoch": 56.47,
+      "grad_norm": 0.01387355849146843,
+      "learning_rate": 4.962938005390836e-06,
+      "loss": 0.0003,
+      "step": 3840
+    },
+    {
+      "epoch": 56.62,
+      "grad_norm": 0.01750769093632698,
+      "learning_rate": 4.946091644204852e-06,
+      "loss": 0.0001,
+      "step": 3850
+    },
+    {
+      "epoch": 56.76,
+      "grad_norm": 0.018308816477656364,
+      "learning_rate": 4.929245283018868e-06,
+      "loss": 0.0001,
+      "step": 3860
+    },
+    {
+      "epoch": 56.91,
+      "grad_norm": 0.0117345554754138,
+      "learning_rate": 4.9123989218328845e-06,
+      "loss": 0.0001,
+      "step": 3870
+    },
+    {
+      "epoch": 57.06,
+      "grad_norm": 0.007436331827193499,
+      "learning_rate": 4.8955525606469e-06,
+      "loss": 0.0001,
+      "step": 3880
+    },
+    {
+      "epoch": 57.21,
+      "grad_norm": 0.00813203677535057,
+      "learning_rate": 4.878706199460917e-06,
+      "loss": 0.0001,
+      "step": 3890
+    },
+    {
+      "epoch": 57.35,
+      "grad_norm": 0.005871808156371117,
+      "learning_rate": 4.8618598382749335e-06,
+      "loss": 0.0001,
+      "step": 3900
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 0.006669571157544851,
+      "learning_rate": 4.845013477088949e-06,
+      "loss": 0.0001,
+      "step": 3910
+    },
+    {
+      "epoch": 57.65,
+      "grad_norm": 0.004995842929929495,
+      "learning_rate": 4.828167115902965e-06,
+      "loss": 0.0001,
+      "step": 3920
+    },
+    {
+      "epoch": 57.79,
+      "grad_norm": 0.0052757407538592815,
+      "learning_rate": 4.811320754716982e-06,
+      "loss": 0.0001,
+      "step": 3930
+    },
+    {
+      "epoch": 57.94,
+      "grad_norm": 0.005870501976460218,
+      "learning_rate": 4.7944743935309975e-06,
+      "loss": 0.0001,
+      "step": 3940
+    },
+    {
+      "epoch": 58.09,
+      "grad_norm": 0.007945972494781017,
+      "learning_rate": 4.777628032345013e-06,
+      "loss": 0.0001,
+      "step": 3950
+    },
+    {
+      "epoch": 58.24,
+      "grad_norm": 0.005463455803692341,
+      "learning_rate": 4.76078167115903e-06,
+      "loss": 0.0001,
+      "step": 3960
+    },
+    {
+      "epoch": 58.38,
+      "grad_norm": 0.0040795220993459225,
+      "learning_rate": 4.7439353099730466e-06,
+      "loss": 0.0001,
+      "step": 3970
+    },
+    {
+      "epoch": 58.53,
+      "grad_norm": 0.004236205480992794,
+      "learning_rate": 4.727088948787062e-06,
+      "loss": 0.0001,
+      "step": 3980
+    },
+    {
+      "epoch": 58.68,
+      "grad_norm": 0.004899207036942244,
+      "learning_rate": 4.710242587601079e-06,
+      "loss": 0.0001,
+      "step": 3990
+    },
+    {
+      "epoch": 58.82,
+      "grad_norm": 0.006042791530489922,
+      "learning_rate": 4.693396226415095e-06,
+      "loss": 0.0001,
+      "step": 4000
+    },
+    {
+      "epoch": 58.82,
+      "eval_loss": 0.5202410221099854,
+      "eval_runtime": 391.5224,
+      "eval_samples_per_second": 1.926,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 40.07905138339921,
+      "step": 4000
+    },
+    {
+      "epoch": 58.97,
+      "grad_norm": 0.004714336711913347,
+      "learning_rate": 4.6765498652291105e-06,
+      "loss": 0.0001,
+      "step": 4010
+    },
+    {
+      "epoch": 59.12,
+      "grad_norm": 0.004793087020516396,
+      "learning_rate": 4.659703504043127e-06,
+      "loss": 0.0001,
+      "step": 4020
+    },
+    {
+      "epoch": 59.26,
+      "grad_norm": 0.0044836923480033875,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.0001,
+      "step": 4030
+    },
+    {
+      "epoch": 59.41,
+      "grad_norm": 0.004963328130543232,
+      "learning_rate": 4.626010781671159e-06,
+      "loss": 0.0001,
+      "step": 4040
+    },
+    {
+      "epoch": 59.56,
+      "grad_norm": 0.003358585759997368,
+      "learning_rate": 4.609164420485175e-06,
+      "loss": 0.0001,
+      "step": 4050
+    },
+    {
+      "epoch": 59.71,
+      "grad_norm": 0.004264355171471834,
+      "learning_rate": 4.592318059299192e-06,
+      "loss": 0.0001,
+      "step": 4060
+    },
+    {
+      "epoch": 59.85,
+      "grad_norm": 0.0040533156134188175,
+      "learning_rate": 4.575471698113208e-06,
+      "loss": 0.0001,
+      "step": 4070
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 0.004796077497303486,
+      "learning_rate": 4.558625336927224e-06,
+      "loss": 0.0001,
+      "step": 4080
+    },
+    {
+      "epoch": 60.15,
+      "grad_norm": 0.0041580419056117535,
+      "learning_rate": 4.54177897574124e-06,
+      "loss": 0.0001,
+      "step": 4090
+    },
+    {
+      "epoch": 60.29,
+      "grad_norm": 0.004052949137985706,
+      "learning_rate": 4.524932614555256e-06,
+      "loss": 0.0,
+      "step": 4100
+    },
+    {
+      "epoch": 60.44,
+      "grad_norm": 0.004524200223386288,
+      "learning_rate": 4.5080862533692726e-06,
+      "loss": 0.0,
+      "step": 4110
+    },
+    {
+      "epoch": 60.59,
+      "grad_norm": 0.0040616546757519245,
+      "learning_rate": 4.491239892183289e-06,
+      "loss": 0.0,
+      "step": 4120
+    },
+    {
+      "epoch": 60.74,
+      "grad_norm": 0.004815140273422003,
+      "learning_rate": 4.474393530997305e-06,
+      "loss": 0.0001,
+      "step": 4130
+    },
+    {
+      "epoch": 60.88,
+      "grad_norm": 0.0043374099768698215,
+      "learning_rate": 4.457547169811321e-06,
+      "loss": 0.0,
+      "step": 4140
+    },
+    {
+      "epoch": 61.03,
+      "grad_norm": 0.003582689445465803,
+      "learning_rate": 4.440700808625337e-06,
+      "loss": 0.0,
+      "step": 4150
+    },
+    {
+      "epoch": 61.18,
+      "grad_norm": 0.00396451773121953,
+      "learning_rate": 4.423854447439353e-06,
+      "loss": 0.0,
+      "step": 4160
+    },
+    {
+      "epoch": 61.32,
+      "grad_norm": 0.004014734644442797,
+      "learning_rate": 4.40700808625337e-06,
+      "loss": 0.0,
+      "step": 4170
+    },
+    {
+      "epoch": 61.47,
+      "grad_norm": 0.0032315212301909924,
+      "learning_rate": 4.390161725067386e-06,
+      "loss": 0.0,
+      "step": 4180
+    },
+    {
+      "epoch": 61.62,
+      "grad_norm": 0.0032560701947659254,
+      "learning_rate": 4.373315363881402e-06,
+      "loss": 0.0001,
+      "step": 4190
+    },
+    {
+      "epoch": 61.76,
+      "grad_norm": 0.0036136566195636988,
+      "learning_rate": 4.356469002695418e-06,
+      "loss": 0.0,
+      "step": 4200
+    },
+    {
+      "epoch": 61.91,
+      "grad_norm": 0.004307006951421499,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 0.0,
+      "step": 4210
+    },
+    {
+      "epoch": 62.06,
+      "grad_norm": 0.00355698075145483,
+      "learning_rate": 4.32277628032345e-06,
+      "loss": 0.0,
+      "step": 4220
+    },
+    {
+      "epoch": 62.21,
+      "grad_norm": 0.003031729022040963,
+      "learning_rate": 4.305929919137466e-06,
+      "loss": 0.0,
+      "step": 4230
+    },
+    {
+      "epoch": 62.35,
+      "grad_norm": 0.003886875230818987,
+      "learning_rate": 4.289083557951483e-06,
+      "loss": 0.0,
+      "step": 4240
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 0.003472729818895459,
+      "learning_rate": 4.2722371967654994e-06,
+      "loss": 0.0,
+      "step": 4250
+    },
+    {
+      "epoch": 62.65,
+      "grad_norm": 0.004264041781425476,
+      "learning_rate": 4.255390835579515e-06,
+      "loss": 0.0,
+      "step": 4260
+    },
+    {
+      "epoch": 62.79,
+      "grad_norm": 0.0033135716803371906,
+      "learning_rate": 4.238544474393531e-06,
+      "loss": 0.0,
+      "step": 4270
+    },
+    {
+      "epoch": 62.94,
+      "grad_norm": 0.003554994473233819,
+      "learning_rate": 4.221698113207548e-06,
+      "loss": 0.0,
+      "step": 4280
+    },
+    {
+      "epoch": 63.09,
+      "grad_norm": 0.0036291780415922403,
+      "learning_rate": 4.204851752021563e-06,
+      "loss": 0.0,
+      "step": 4290
+    },
+    {
+      "epoch": 63.24,
+      "grad_norm": 0.003123074769973755,
+      "learning_rate": 4.18800539083558e-06,
+      "loss": 0.0,
+      "step": 4300
+    },
+    {
+      "epoch": 63.38,
+      "grad_norm": 0.0030143638141453266,
+      "learning_rate": 4.171159029649597e-06,
+      "loss": 0.0,
+      "step": 4310
+    },
+    {
+      "epoch": 63.53,
+      "grad_norm": 0.0033236437011510134,
+      "learning_rate": 4.1543126684636125e-06,
+      "loss": 0.0,
+      "step": 4320
+    },
+    {
+      "epoch": 63.68,
+      "grad_norm": 0.003377540735527873,
+      "learning_rate": 4.137466307277628e-06,
+      "loss": 0.0,
+      "step": 4330
+    },
+    {
+      "epoch": 63.82,
+      "grad_norm": 0.0034374604001641273,
+      "learning_rate": 4.120619946091645e-06,
+      "loss": 0.0,
+      "step": 4340
+    },
+    {
+      "epoch": 63.97,
+      "grad_norm": 0.0032602883875370026,
+      "learning_rate": 4.103773584905661e-06,
+      "loss": 0.0,
+      "step": 4350
+    },
+    {
+      "epoch": 64.12,
+      "grad_norm": 0.003039925592020154,
+      "learning_rate": 4.086927223719676e-06,
+      "loss": 0.0,
+      "step": 4360
+    },
+    {
+      "epoch": 64.26,
+      "grad_norm": 0.003154992824420333,
+      "learning_rate": 4.070080862533693e-06,
+      "loss": 0.0,
+      "step": 4370
+    },
+    {
+      "epoch": 64.41,
+      "grad_norm": 0.0030352873727679253,
+      "learning_rate": 4.053234501347709e-06,
+      "loss": 0.0,
+      "step": 4380
+    },
+    {
+      "epoch": 64.56,
+      "grad_norm": 0.002541678724810481,
+      "learning_rate": 4.0363881401617255e-06,
+      "loss": 0.0,
+      "step": 4390
+    },
+    {
+      "epoch": 64.71,
+      "grad_norm": 0.0035515627823770046,
+      "learning_rate": 4.019541778975742e-06,
+      "loss": 0.0,
+      "step": 4400
+    },
+    {
+      "epoch": 64.85,
+      "grad_norm": 0.0032643494196236134,
+      "learning_rate": 4.002695417789758e-06,
+      "loss": 0.0,
+      "step": 4410
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 0.0029830934945493937,
+      "learning_rate": 3.985849056603774e-06,
+      "loss": 0.0,
+      "step": 4420
+    },
+    {
+      "epoch": 65.15,
+      "grad_norm": 0.003434343496337533,
+      "learning_rate": 3.96900269541779e-06,
+      "loss": 0.0,
+      "step": 4430
+    },
+    {
+      "epoch": 65.29,
+      "grad_norm": 0.003297910327091813,
+      "learning_rate": 3.952156334231806e-06,
+      "loss": 0.0,
+      "step": 4440
+    },
+    {
+      "epoch": 65.44,
+      "grad_norm": 0.0029679101426154375,
+      "learning_rate": 3.935309973045822e-06,
+      "loss": 0.0,
+      "step": 4450
+    },
+    {
+      "epoch": 65.59,
+      "grad_norm": 0.002996984403580427,
+      "learning_rate": 3.9184636118598385e-06,
+      "loss": 0.0,
+      "step": 4460
+    },
+    {
+      "epoch": 65.74,
+      "grad_norm": 0.0031057405285537243,
+      "learning_rate": 3.901617250673855e-06,
+      "loss": 0.0,
+      "step": 4470
+    },
+    {
+      "epoch": 65.88,
+      "grad_norm": 0.0026742741465568542,
+      "learning_rate": 3.884770889487871e-06,
+      "loss": 0.0,
+      "step": 4480
+    },
+    {
+      "epoch": 66.03,
+      "grad_norm": 0.003236331744119525,
+      "learning_rate": 3.8679245283018875e-06,
+      "loss": 0.0,
+      "step": 4490
+    },
+    {
+      "epoch": 66.18,
+      "grad_norm": 0.002975077135488391,
+      "learning_rate": 3.851078167115903e-06,
+      "loss": 0.0,
+      "step": 4500
+    },
+    {
+      "epoch": 66.18,
+      "eval_loss": 0.5316298604011536,
+      "eval_runtime": 394.5635,
+      "eval_samples_per_second": 1.911,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 39.80237154150198,
+      "step": 4500
+    },
+    {
+      "epoch": 66.32,
+      "grad_norm": 0.0026845140382647514,
+      "learning_rate": 3.834231805929919e-06,
+      "loss": 0.0,
+      "step": 4510
+    },
+    {
+      "epoch": 66.47,
+      "grad_norm": 0.002751615596935153,
+      "learning_rate": 3.817385444743936e-06,
+      "loss": 0.0,
+      "step": 4520
+    },
+    {
+      "epoch": 66.62,
+      "grad_norm": 0.0025983687955886126,
+      "learning_rate": 3.800539083557952e-06,
+      "loss": 0.0,
+      "step": 4530
+    },
+    {
+      "epoch": 66.76,
+      "grad_norm": 0.0027817366644740105,
+      "learning_rate": 3.7836927223719677e-06,
+      "loss": 0.0,
+      "step": 4540
+    },
+    {
+      "epoch": 66.91,
+      "grad_norm": 0.0028080460615456104,
+      "learning_rate": 3.766846361185984e-06,
+      "loss": 0.0,
+      "step": 4550
+    },
+    {
+      "epoch": 67.06,
+      "grad_norm": 0.0026772802229970694,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.0,
+      "step": 4560
+    },
+    {
+      "epoch": 67.21,
+      "grad_norm": 0.002854007761925459,
+      "learning_rate": 3.7331536388140167e-06,
+      "loss": 0.0,
+      "step": 4570
+    },
+    {
+      "epoch": 67.35,
+      "grad_norm": 0.0026527883019298315,
+      "learning_rate": 3.716307277628033e-06,
+      "loss": 0.0,
+      "step": 4580
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 0.002448031213134527,
+      "learning_rate": 3.6994609164420487e-06,
+      "loss": 0.0,
+      "step": 4590
+    },
+    {
+      "epoch": 67.65,
+      "grad_norm": 0.002617811318486929,
+      "learning_rate": 3.682614555256065e-06,
+      "loss": 0.0,
+      "step": 4600
+    },
+    {
+      "epoch": 67.79,
+      "grad_norm": 0.0031759808771312237,
+      "learning_rate": 3.665768194070081e-06,
+      "loss": 0.0,
+      "step": 4610
+    },
+    {
+      "epoch": 67.94,
+      "grad_norm": 0.002507594181224704,
+      "learning_rate": 3.6489218328840973e-06,
+      "loss": 0.0,
+      "step": 4620
+    },
+    {
+      "epoch": 68.09,
+      "grad_norm": 0.0023514223285019398,
+      "learning_rate": 3.632075471698113e-06,
+      "loss": 0.0,
+      "step": 4630
+    },
+    {
+      "epoch": 68.24,
+      "grad_norm": 0.0027671242132782936,
+      "learning_rate": 3.6152291105121297e-06,
+      "loss": 0.0,
+      "step": 4640
+    },
+    {
+      "epoch": 68.38,
+      "grad_norm": 0.002477505709975958,
+      "learning_rate": 3.598382749326146e-06,
+      "loss": 0.0,
+      "step": 4650
+    },
+    {
+      "epoch": 68.53,
+      "grad_norm": 0.0028106619138270617,
+      "learning_rate": 3.581536388140162e-06,
+      "loss": 0.0,
+      "step": 4660
+    },
+    {
+      "epoch": 68.68,
+      "grad_norm": 0.0034189883153885603,
+      "learning_rate": 3.5646900269541783e-06,
+      "loss": 0.0,
+      "step": 4670
+    },
+    {
+      "epoch": 68.82,
+      "grad_norm": 0.0030117544811218977,
+      "learning_rate": 3.547843665768194e-06,
+      "loss": 0.0,
+      "step": 4680
+    },
+    {
+      "epoch": 68.97,
+      "grad_norm": 0.002722077304497361,
+      "learning_rate": 3.5309973045822103e-06,
+      "loss": 0.0,
+      "step": 4690
+    },
+    {
+      "epoch": 69.12,
+      "grad_norm": 0.00265983073040843,
+      "learning_rate": 3.514150943396227e-06,
+      "loss": 0.0,
+      "step": 4700
+    },
+    {
+      "epoch": 69.26,
+      "grad_norm": 0.002687458647415042,
+      "learning_rate": 3.497304582210243e-06,
+      "loss": 0.0,
+      "step": 4710
+    },
+    {
+      "epoch": 69.41,
+      "grad_norm": 0.002611985895782709,
+      "learning_rate": 3.480458221024259e-06,
+      "loss": 0.0,
+      "step": 4720
+    },
+    {
+      "epoch": 69.56,
+      "grad_norm": 0.002745421137660742,
+      "learning_rate": 3.463611859838275e-06,
+      "loss": 0.0,
+      "step": 4730
+    },
+    {
+      "epoch": 69.71,
+      "grad_norm": 0.00274544395506382,
+      "learning_rate": 3.4467654986522914e-06,
+      "loss": 0.0,
+      "step": 4740
+    },
+    {
+      "epoch": 69.85,
+      "grad_norm": 0.0032134081702679396,
+      "learning_rate": 3.4299191374663076e-06,
+      "loss": 0.0,
+      "step": 4750
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 0.0027218926697969437,
+      "learning_rate": 3.4130727762803238e-06,
+      "loss": 0.0,
+      "step": 4760
+    },
+    {
+      "epoch": 70.15,
+      "grad_norm": 0.0029763532802462578,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 0.0,
+      "step": 4770
+    },
+    {
+      "epoch": 70.29,
+      "grad_norm": 0.002971038920804858,
+      "learning_rate": 3.379380053908356e-06,
+      "loss": 0.0,
+      "step": 4780
+    },
+    {
+      "epoch": 70.44,
+      "grad_norm": 0.002593262353911996,
+      "learning_rate": 3.3625336927223724e-06,
+      "loss": 0.0,
+      "step": 4790
+    },
+    {
+      "epoch": 70.59,
+      "grad_norm": 0.0025443334598094225,
+      "learning_rate": 3.3456873315363886e-06,
+      "loss": 0.0,
+      "step": 4800
+    },
+    {
+      "epoch": 70.74,
+      "grad_norm": 0.002275052247568965,
+      "learning_rate": 3.3288409703504044e-06,
+      "loss": 0.0,
+      "step": 4810
+    },
+    {
+      "epoch": 70.88,
+      "grad_norm": 0.0028011808171868324,
+      "learning_rate": 3.3119946091644206e-06,
+      "loss": 0.0,
+      "step": 4820
+    },
+    {
+      "epoch": 71.03,
+      "grad_norm": 0.0028713145293295383,
+      "learning_rate": 3.2951482479784368e-06,
+      "loss": 0.0,
+      "step": 4830
+    },
+    {
+      "epoch": 71.18,
+      "grad_norm": 0.00250536622479558,
+      "learning_rate": 3.2783018867924534e-06,
+      "loss": 0.0,
+      "step": 4840
+    },
+    {
+      "epoch": 71.32,
+      "grad_norm": 0.002349557587876916,
+      "learning_rate": 3.2614555256064696e-06,
+      "loss": 0.0,
+      "step": 4850
+    },
+    {
+      "epoch": 71.47,
+      "grad_norm": 0.002672821283340454,
+      "learning_rate": 3.2446091644204854e-06,
+      "loss": 0.0,
+      "step": 4860
+    },
+    {
+      "epoch": 71.62,
+      "grad_norm": 0.002757046138867736,
+      "learning_rate": 3.2277628032345016e-06,
+      "loss": 0.0,
+      "step": 4870
+    },
+    {
+      "epoch": 71.76,
+      "grad_norm": 0.002534884260967374,
+      "learning_rate": 3.210916442048518e-06,
+      "loss": 0.0,
+      "step": 4880
+    },
+    {
+      "epoch": 71.91,
+      "grad_norm": 0.0023479722440242767,
+      "learning_rate": 3.194070080862534e-06,
+      "loss": 0.0,
+      "step": 4890
+    },
+    {
+      "epoch": 72.06,
+      "grad_norm": 0.0025629138108342886,
+      "learning_rate": 3.1772237196765498e-06,
+      "loss": 0.0,
+      "step": 4900
+    },
+    {
+      "epoch": 72.21,
+      "grad_norm": 0.002501903334632516,
+      "learning_rate": 3.160377358490566e-06,
+      "loss": 0.0,
+      "step": 4910
+    },
+    {
+      "epoch": 72.35,
+      "grad_norm": 0.002238241257146001,
+      "learning_rate": 3.1435309973045826e-06,
+      "loss": 0.0,
+      "step": 4920
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 0.0019617443904280663,
+      "learning_rate": 3.126684636118599e-06,
+      "loss": 0.0,
+      "step": 4930
+    },
+    {
+      "epoch": 72.65,
+      "grad_norm": 0.0025888625532388687,
+      "learning_rate": 3.109838274932615e-06,
+      "loss": 0.0,
+      "step": 4940
+    },
+    {
+      "epoch": 72.79,
+      "grad_norm": 0.0022458680905401707,
+      "learning_rate": 3.092991913746631e-06,
+      "loss": 0.0,
+      "step": 4950
+    },
+    {
+      "epoch": 72.94,
+      "grad_norm": 0.0028502140194177628,
+      "learning_rate": 3.076145552560647e-06,
+      "loss": 0.0,
+      "step": 4960
+    },
+    {
+      "epoch": 73.09,
+      "grad_norm": 0.0024815101642161608,
+      "learning_rate": 3.0592991913746632e-06,
+      "loss": 0.0,
+      "step": 4970
+    },
+    {
+      "epoch": 73.24,
+      "grad_norm": 0.0020565763115882874,
+      "learning_rate": 3.04245283018868e-06,
+      "loss": 0.0,
+      "step": 4980
+    },
+    {
+      "epoch": 73.38,
+      "grad_norm": 0.0019868898671120405,
+      "learning_rate": 3.025606469002695e-06,
+      "loss": 0.0,
+      "step": 4990
+    },
+    {
+      "epoch": 73.53,
+      "grad_norm": 0.002612064126878977,
+      "learning_rate": 3.008760107816712e-06,
+      "loss": 0.0,
+      "step": 5000
+    },
+    {
+      "epoch": 73.53,
+      "eval_loss": 0.5392932891845703,
+      "eval_runtime": 438.3941,
+      "eval_samples_per_second": 1.72,
+      "eval_steps_per_second": 0.018,
+      "eval_wer": 39.901185770750985,
+      "step": 5000
+    },
+    {
+      "epoch": 73.68,
+      "grad_norm": 0.0019985686521977186,
+      "learning_rate": 2.991913746630728e-06,
+      "loss": 0.0,
+      "step": 5010
+    },
+    {
+      "epoch": 73.82,
+      "grad_norm": 0.0021473197266459465,
+      "learning_rate": 2.9750673854447442e-06,
+      "loss": 0.0,
+      "step": 5020
+    },
+    {
+      "epoch": 73.97,
+      "grad_norm": 0.0023225159384310246,
+      "learning_rate": 2.9582210242587605e-06,
+      "loss": 0.0,
+      "step": 5030
+    },
+    {
+      "epoch": 74.12,
+      "grad_norm": 0.0022743339650332928,
+      "learning_rate": 2.9413746630727762e-06,
+      "loss": 0.0,
+      "step": 5040
+    },
+    {
+      "epoch": 74.26,
+      "grad_norm": 0.002134987385943532,
+      "learning_rate": 2.9245283018867924e-06,
+      "loss": 0.0,
+      "step": 5050
+    },
+    {
+      "epoch": 74.41,
+      "grad_norm": 0.0021126819774508476,
+      "learning_rate": 2.907681940700809e-06,
+      "loss": 0.0,
+      "step": 5060
+    },
+    {
+      "epoch": 74.56,
+      "grad_norm": 0.002264765091240406,
+      "learning_rate": 2.8908355795148253e-06,
+      "loss": 0.0,
+      "step": 5070
+    },
+    {
+      "epoch": 74.71,
+      "grad_norm": 0.0028144866228103638,
+      "learning_rate": 2.8739892183288415e-06,
+      "loss": 0.0,
+      "step": 5080
+    },
+    {
+      "epoch": 74.85,
+      "grad_norm": 0.0017289798706769943,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.0,
+      "step": 5090
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 0.00249221152625978,
+      "learning_rate": 2.8402964959568735e-06,
+      "loss": 0.0,
+      "step": 5100
+    },
+    {
+      "epoch": 75.15,
+      "grad_norm": 0.0018814760260283947,
+      "learning_rate": 2.8234501347708897e-06,
+      "loss": 0.0,
+      "step": 5110
+    },
+    {
+      "epoch": 75.29,
+      "grad_norm": 0.0022003494668751955,
+      "learning_rate": 2.8066037735849063e-06,
+      "loss": 0.0,
+      "step": 5120
+    },
+    {
+      "epoch": 75.44,
+      "grad_norm": 0.0023006245028227568,
+      "learning_rate": 2.7897574123989217e-06,
+      "loss": 0.0,
+      "step": 5130
+    },
+    {
+      "epoch": 75.59,
+      "grad_norm": 0.0018226418178528547,
+      "learning_rate": 2.7729110512129383e-06,
+      "loss": 0.0,
+      "step": 5140
+    },
+    {
+      "epoch": 75.74,
+      "grad_norm": 0.0019806609489023685,
+      "learning_rate": 2.7560646900269545e-06,
+      "loss": 0.0,
+      "step": 5150
+    },
+    {
+      "epoch": 75.88,
+      "grad_norm": 0.0021330551244318485,
+      "learning_rate": 2.7392183288409707e-06,
+      "loss": 0.0,
+      "step": 5160
+    },
+    {
+      "epoch": 76.03,
+      "grad_norm": 0.0021852573845535517,
+      "learning_rate": 2.722371967654987e-06,
+      "loss": 0.0,
+      "step": 5170
+    },
+    {
+      "epoch": 76.18,
+      "grad_norm": 0.0021261894144117832,
+      "learning_rate": 2.7055256064690027e-06,
+      "loss": 0.0,
+      "step": 5180
+    },
+    {
+      "epoch": 76.32,
+      "grad_norm": 0.0022390589583665133,
+      "learning_rate": 2.688679245283019e-06,
+      "loss": 0.0,
+      "step": 5190
+    },
+    {
+      "epoch": 76.47,
+      "grad_norm": 0.002223338931798935,
+      "learning_rate": 2.6718328840970355e-06,
+      "loss": 0.0,
+      "step": 5200
+    },
+    {
+      "epoch": 76.62,
+      "grad_norm": 0.0019250946352258325,
+      "learning_rate": 2.6549865229110517e-06,
+      "loss": 0.0,
+      "step": 5210
+    },
+    {
+      "epoch": 76.76,
+      "grad_norm": 0.001935865031555295,
+      "learning_rate": 2.6381401617250675e-06,
+      "loss": 0.0,
+      "step": 5220
+    },
+    {
+      "epoch": 76.91,
+      "grad_norm": 0.0018222652142867446,
+      "learning_rate": 2.6212938005390837e-06,
+      "loss": 0.0,
+      "step": 5230
+    },
+    {
+      "epoch": 77.06,
+      "grad_norm": 0.0015805804869160056,
+      "learning_rate": 2.6044474393531e-06,
+      "loss": 0.0,
+      "step": 5240
+    },
+    {
+      "epoch": 77.21,
+      "grad_norm": 0.0018061785958707333,
+      "learning_rate": 2.587601078167116e-06,
+      "loss": 0.0,
+      "step": 5250
+    },
+    {
+      "epoch": 77.35,
+      "grad_norm": 0.0019492105348035693,
+      "learning_rate": 2.5707547169811327e-06,
+      "loss": 0.0,
+      "step": 5260
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 0.0020225332118570805,
+      "learning_rate": 2.553908355795148e-06,
+      "loss": 0.0,
+      "step": 5270
+    },
+    {
+      "epoch": 77.65,
+      "grad_norm": 0.0021102512255311012,
+      "learning_rate": 2.5370619946091647e-06,
+      "loss": 0.0,
+      "step": 5280
+    },
+    {
+      "epoch": 77.79,
+      "grad_norm": 0.001952395774424076,
+      "learning_rate": 2.520215633423181e-06,
+      "loss": 0.0,
+      "step": 5290
+    },
+    {
+      "epoch": 77.94,
+      "grad_norm": 0.0022354356478899717,
+      "learning_rate": 2.503369272237197e-06,
+      "loss": 0.0,
+      "step": 5300
+    },
+    {
+      "epoch": 78.09,
+      "grad_norm": 0.0015943964244797826,
+      "learning_rate": 2.4865229110512133e-06,
+      "loss": 0.0,
+      "step": 5310
+    },
+    {
+      "epoch": 78.24,
+      "grad_norm": 0.00202794186770916,
+      "learning_rate": 2.469676549865229e-06,
+      "loss": 0.0,
+      "step": 5320
+    },
+    {
+      "epoch": 78.38,
+      "grad_norm": 0.002233139704912901,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 0.0,
+      "step": 5330
+    },
+    {
+      "epoch": 78.53,
+      "grad_norm": 0.0019945132080465555,
+      "learning_rate": 2.435983827493262e-06,
+      "loss": 0.0,
+      "step": 5340
+    },
+    {
+      "epoch": 78.68,
+      "grad_norm": 0.001916647655889392,
+      "learning_rate": 2.4191374663072777e-06,
+      "loss": 0.0,
+      "step": 5350
+    },
+    {
+      "epoch": 78.82,
+      "grad_norm": 0.0020198444835841656,
+      "learning_rate": 2.402291105121294e-06,
+      "loss": 0.0,
+      "step": 5360
+    },
+    {
+      "epoch": 78.97,
+      "grad_norm": 0.0019901052583009005,
+      "learning_rate": 2.38544474393531e-06,
+      "loss": 0.0,
+      "step": 5370
+    },
+    {
+      "epoch": 79.12,
+      "grad_norm": 0.0020235173869878054,
+      "learning_rate": 2.3685983827493263e-06,
+      "loss": 0.0,
+      "step": 5380
+    },
+    {
+      "epoch": 79.26,
+      "grad_norm": 0.0018330395687371492,
+      "learning_rate": 2.3517520215633426e-06,
+      "loss": 0.0,
+      "step": 5390
+    },
+    {
+      "epoch": 79.41,
+      "grad_norm": 0.0023276926949620247,
+      "learning_rate": 2.3349056603773588e-06,
+      "loss": 0.0,
+      "step": 5400
+    },
+    {
+      "epoch": 79.56,
+      "grad_norm": 0.0019117222400382161,
+      "learning_rate": 2.3180592991913745e-06,
+      "loss": 0.0,
+      "step": 5410
+    },
+    {
+      "epoch": 79.71,
+      "grad_norm": 0.0019178889924660325,
+      "learning_rate": 2.301212938005391e-06,
+      "loss": 0.0,
+      "step": 5420
+    },
+    {
+      "epoch": 79.85,
+      "grad_norm": 0.002448749728500843,
+      "learning_rate": 2.2843665768194074e-06,
+      "loss": 0.0,
+      "step": 5430
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 0.0022572469897568226,
+      "learning_rate": 2.267520215633423e-06,
+      "loss": 0.0,
+      "step": 5440
+    },
+    {
+      "epoch": 80.15,
+      "grad_norm": 0.001967059215530753,
+      "learning_rate": 2.2506738544474398e-06,
+      "loss": 0.0,
+      "step": 5450
+    },
+    {
+      "epoch": 80.29,
+      "grad_norm": 0.0022357290145009756,
+      "learning_rate": 2.2338274932614556e-06,
+      "loss": 0.0,
+      "step": 5460
+    },
+    {
+      "epoch": 80.44,
+      "grad_norm": 0.002172647975385189,
+      "learning_rate": 2.2169811320754718e-06,
+      "loss": 0.0,
+      "step": 5470
+    },
+    {
+      "epoch": 80.59,
+      "grad_norm": 0.0016459524631500244,
+      "learning_rate": 2.200134770889488e-06,
+      "loss": 0.0,
+      "step": 5480
+    },
+    {
+      "epoch": 80.74,
+      "grad_norm": 0.00204639439471066,
+      "learning_rate": 2.183288409703504e-06,
+      "loss": 0.0,
+      "step": 5490
+    },
+    {
+      "epoch": 80.88,
+      "grad_norm": 0.0017498856177553535,
+      "learning_rate": 2.1664420485175204e-06,
+      "loss": 0.0,
+      "step": 5500
+    },
+    {
+      "epoch": 80.88,
+      "eval_loss": 0.5448271632194519,
+      "eval_runtime": 430.4314,
+      "eval_samples_per_second": 1.752,
+      "eval_steps_per_second": 0.019,
+      "eval_wer": 39.66403162055336,
+      "step": 5500
+    },
+    {
+      "epoch": 81.03,
+      "grad_norm": 0.0018372549675405025,
+      "learning_rate": 2.1495956873315366e-06,
+      "loss": 0.0,
+      "step": 5510
+    },
+    {
+      "epoch": 81.18,
+      "grad_norm": 0.0016495827585458755,
+      "learning_rate": 2.132749326145553e-06,
+      "loss": 0.0,
+      "step": 5520
+    },
+    {
+      "epoch": 81.32,
+      "grad_norm": 0.0017999428091570735,
+      "learning_rate": 2.115902964959569e-06,
+      "loss": 0.0,
+      "step": 5530
+    },
+    {
+      "epoch": 81.47,
+      "grad_norm": 0.0018734214827418327,
+      "learning_rate": 2.099056603773585e-06,
+      "loss": 0.0,
+      "step": 5540
+    },
+    {
+      "epoch": 81.62,
+      "grad_norm": 0.0016185399144887924,
+      "learning_rate": 2.082210242587601e-06,
+      "loss": 0.0,
+      "step": 5550
+    },
+    {
+      "epoch": 81.76,
+      "grad_norm": 0.001710501266643405,
+      "learning_rate": 2.0653638814016176e-06,
+      "loss": 0.0,
+      "step": 5560
+    },
+    {
+      "epoch": 81.91,
+      "grad_norm": 0.002252147998660803,
+      "learning_rate": 2.0485175202156334e-06,
+      "loss": 0.0,
+      "step": 5570
+    },
+    {
+      "epoch": 82.06,
+      "grad_norm": 0.0018945408519357443,
+      "learning_rate": 2.0316711590296496e-06,
+      "loss": 0.0,
+      "step": 5580
+    },
+    {
+      "epoch": 82.21,
+      "grad_norm": 0.0016931135905906558,
+      "learning_rate": 2.014824797843666e-06,
+      "loss": 0.0,
+      "step": 5590
+    },
+    {
+      "epoch": 82.35,
+      "grad_norm": 0.002079603960737586,
+      "learning_rate": 1.997978436657682e-06,
+      "loss": 0.0,
+      "step": 5600
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 0.0014379196800291538,
+      "learning_rate": 1.981132075471698e-06,
+      "loss": 0.0,
+      "step": 5610
+    },
+    {
+      "epoch": 82.65,
+      "grad_norm": 0.0017873428296297789,
+      "learning_rate": 1.9642857142857144e-06,
+      "loss": 0.0,
+      "step": 5620
+    },
+    {
+      "epoch": 82.79,
+      "grad_norm": 0.0016607646830379963,
+      "learning_rate": 1.9474393530997306e-06,
+      "loss": 0.0,
+      "step": 5630
+    },
+    {
+      "epoch": 82.94,
+      "grad_norm": 0.0018578683957457542,
+      "learning_rate": 1.930592991913747e-06,
+      "loss": 0.0,
+      "step": 5640
+    },
+    {
+      "epoch": 83.09,
+      "grad_norm": 0.001992453821003437,
+      "learning_rate": 1.913746630727763e-06,
+      "loss": 0.0,
+      "step": 5650
+    },
+    {
+      "epoch": 83.24,
+      "grad_norm": 0.001795714139007032,
+      "learning_rate": 1.896900269541779e-06,
+      "loss": 0.0,
+      "step": 5660
+    },
+    {
+      "epoch": 83.38,
+      "grad_norm": 0.0017513408092781901,
+      "learning_rate": 1.8800539083557952e-06,
+      "loss": 0.0,
+      "step": 5670
+    },
+    {
+      "epoch": 83.53,
+      "grad_norm": 0.001984973670914769,
+      "learning_rate": 1.8632075471698114e-06,
+      "loss": 0.0,
+      "step": 5680
+    },
+    {
+      "epoch": 83.68,
+      "grad_norm": 0.001725591835565865,
+      "learning_rate": 1.8463611859838276e-06,
+      "loss": 0.0,
+      "step": 5690
+    },
+    {
+      "epoch": 83.82,
+      "grad_norm": 0.001957892207428813,
+      "learning_rate": 1.8295148247978438e-06,
+      "loss": 0.0,
+      "step": 5700
+    },
+    {
+      "epoch": 83.97,
+      "grad_norm": 0.0014876240165904164,
+      "learning_rate": 1.8126684636118598e-06,
+      "loss": 0.0,
+      "step": 5710
+    },
+    {
+      "epoch": 84.12,
+      "grad_norm": 0.0017114444635808468,
+      "learning_rate": 1.7958221024258763e-06,
+      "loss": 0.0,
+      "step": 5720
+    },
+    {
+      "epoch": 84.26,
+      "grad_norm": 0.0018760694656521082,
+      "learning_rate": 1.7789757412398922e-06,
+      "loss": 0.0,
+      "step": 5730
+    },
+    {
+      "epoch": 84.41,
+      "grad_norm": 0.001803001156076789,
+      "learning_rate": 1.7621293800539085e-06,
+      "loss": 0.0,
+      "step": 5740
+    },
+    {
+      "epoch": 84.56,
+      "grad_norm": 0.0018211943097412586,
+      "learning_rate": 1.7452830188679247e-06,
+      "loss": 0.0,
+      "step": 5750
+    },
+    {
+      "epoch": 84.71,
+      "grad_norm": 0.001698785461485386,
+      "learning_rate": 1.7284366576819409e-06,
+      "loss": 0.0,
+      "step": 5760
+    },
+    {
+      "epoch": 84.85,
+      "grad_norm": 0.00188141327816993,
+      "learning_rate": 1.711590296495957e-06,
+      "loss": 0.0,
+      "step": 5770
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 0.0018102971371263266,
+      "learning_rate": 1.6947439353099733e-06,
+      "loss": 0.0,
+      "step": 5780
+    },
+    {
+      "epoch": 85.15,
+      "grad_norm": 0.002033288823440671,
+      "learning_rate": 1.6778975741239895e-06,
+      "loss": 0.0,
+      "step": 5790
+    },
+    {
+      "epoch": 85.29,
+      "grad_norm": 0.0015100777382031083,
+      "learning_rate": 1.6610512129380055e-06,
+      "loss": 0.0,
+      "step": 5800
+    },
+    {
+      "epoch": 85.44,
+      "grad_norm": 0.0018259905045852065,
+      "learning_rate": 1.6442048517520217e-06,
+      "loss": 0.0,
+      "step": 5810
+    },
+    {
+      "epoch": 85.59,
+      "grad_norm": 0.0018028096528723836,
+      "learning_rate": 1.6273584905660379e-06,
+      "loss": 0.0,
+      "step": 5820
+    },
+    {
+      "epoch": 85.74,
+      "grad_norm": 0.0016898865578696132,
+      "learning_rate": 1.610512129380054e-06,
+      "loss": 0.0,
+      "step": 5830
+    },
+    {
+      "epoch": 85.88,
+      "grad_norm": 0.0020932299084961414,
+      "learning_rate": 1.59366576819407e-06,
+      "loss": 0.0,
+      "step": 5840
+    },
+    {
+      "epoch": 86.03,
+      "grad_norm": 0.0018589514074847102,
+      "learning_rate": 1.5768194070080865e-06,
+      "loss": 0.0,
+      "step": 5850
+    },
+    {
+      "epoch": 86.18,
+      "grad_norm": 0.001960631925612688,
+      "learning_rate": 1.5599730458221027e-06,
+      "loss": 0.0,
+      "step": 5860
+    },
+    {
+      "epoch": 86.32,
+      "grad_norm": 0.0015809714095667005,
+      "learning_rate": 1.5431266846361187e-06,
+      "loss": 0.0,
+      "step": 5870
+    },
+    {
+      "epoch": 86.47,
+      "grad_norm": 0.0015852567739784718,
+      "learning_rate": 1.5262803234501349e-06,
+      "loss": 0.0,
+      "step": 5880
+    },
+    {
+      "epoch": 86.62,
+      "grad_norm": 0.001482869847677648,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 0.0,
+      "step": 5890
+    },
+    {
+      "epoch": 86.76,
+      "grad_norm": 0.00168868328910321,
+      "learning_rate": 1.4925876010781673e-06,
+      "loss": 0.0,
+      "step": 5900
+    },
+    {
+      "epoch": 86.91,
+      "grad_norm": 0.0017209333600476384,
+      "learning_rate": 1.4757412398921833e-06,
+      "loss": 0.0,
+      "step": 5910
+    },
+    {
+      "epoch": 87.06,
+      "grad_norm": 0.0015247148694470525,
+      "learning_rate": 1.4588948787061997e-06,
+      "loss": 0.0,
+      "step": 5920
+    },
+    {
+      "epoch": 87.21,
+      "grad_norm": 0.0020303872879594564,
+      "learning_rate": 1.4420485175202157e-06,
+      "loss": 0.0,
+      "step": 5930
+    },
+    {
+      "epoch": 87.35,
+      "grad_norm": 0.001518064527772367,
+      "learning_rate": 1.425202156334232e-06,
+      "loss": 0.0,
+      "step": 5940
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 0.0015475251711905003,
+      "learning_rate": 1.4083557951482481e-06,
+      "loss": 0.0,
+      "step": 5950
+    },
+    {
+      "epoch": 87.65,
+      "grad_norm": 0.0014926763251423836,
+      "learning_rate": 1.3915094339622643e-06,
+      "loss": 0.0,
+      "step": 5960
+    },
+    {
+      "epoch": 87.79,
+      "grad_norm": 0.0014638776192441583,
+      "learning_rate": 1.3746630727762805e-06,
+      "loss": 0.0,
+      "step": 5970
+    },
+    {
+      "epoch": 87.94,
+      "grad_norm": 0.0015593844000250101,
+      "learning_rate": 1.3578167115902965e-06,
+      "loss": 0.0,
+      "step": 5980
+    },
+    {
+      "epoch": 88.09,
+      "grad_norm": 0.001806886401027441,
+      "learning_rate": 1.340970350404313e-06,
+      "loss": 0.0,
+      "step": 5990
+    },
+    {
+      "epoch": 88.24,
+      "grad_norm": 0.001656343461945653,
+      "learning_rate": 1.324123989218329e-06,
+      "loss": 0.0,
+      "step": 6000
+    },
+    {
+      "epoch": 88.24,
+      "eval_loss": 0.5489587783813477,
+      "eval_runtime": 428.0298,
+      "eval_samples_per_second": 1.762,
+      "eval_steps_per_second": 0.019,
+      "eval_wer": 39.565217391304344,
+      "step": 6000
+    },
+    {
+      "epoch": 88.38,
+      "grad_norm": 0.0019978652708232403,
+      "learning_rate": 1.3072776280323451e-06,
+      "loss": 0.0,
+      "step": 6010
+    },
+    {
+      "epoch": 88.53,
+      "grad_norm": 0.0016022637719288468,
+      "learning_rate": 1.2904312668463611e-06,
+      "loss": 0.0,
+      "step": 6020
+    },
+    {
+      "epoch": 88.68,
+      "grad_norm": 0.0016824258491396904,
+      "learning_rate": 1.2735849056603775e-06,
+      "loss": 0.0,
+      "step": 6030
+    },
+    {
+      "epoch": 88.82,
+      "grad_norm": 0.0017654020339250565,
+      "learning_rate": 1.2567385444743937e-06,
+      "loss": 0.0,
+      "step": 6040
+    },
+    {
+      "epoch": 88.97,
+      "grad_norm": 0.001638589659705758,
+      "learning_rate": 1.2398921832884097e-06,
+      "loss": 0.0,
+      "step": 6050
+    },
+    {
+      "epoch": 89.12,
+      "grad_norm": 0.001573295914568007,
+      "learning_rate": 1.223045822102426e-06,
+      "loss": 0.0,
+      "step": 6060
+    },
+    {
+      "epoch": 89.26,
+      "grad_norm": 0.0014727225061506033,
+      "learning_rate": 1.2061994609164422e-06,
+      "loss": 0.0,
+      "step": 6070
+    },
+    {
+      "epoch": 89.41,
+      "grad_norm": 0.0016162623651325703,
+      "learning_rate": 1.1893530997304584e-06,
+      "loss": 0.0,
+      "step": 6080
+    },
+    {
+      "epoch": 89.56,
+      "grad_norm": 0.0016934397863224149,
+      "learning_rate": 1.1725067385444746e-06,
+      "loss": 0.0,
+      "step": 6090
+    },
+    {
+      "epoch": 89.71,
+      "grad_norm": 0.0017905846470966935,
+      "learning_rate": 1.1556603773584908e-06,
+      "loss": 0.0,
+      "step": 6100
+    },
+    {
+      "epoch": 89.85,
+      "grad_norm": 0.001646560151129961,
+      "learning_rate": 1.1388140161725068e-06,
+      "loss": 0.0,
+      "step": 6110
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 0.001669664983637631,
+      "learning_rate": 1.121967654986523e-06,
+      "loss": 0.0,
+      "step": 6120
+    },
+    {
+      "epoch": 90.15,
+      "grad_norm": 0.0015495212282985449,
+      "learning_rate": 1.1051212938005392e-06,
+      "loss": 0.0,
+      "step": 6130
+    },
+    {
+      "epoch": 90.29,
+      "grad_norm": 0.0015938329743221402,
+      "learning_rate": 1.0882749326145554e-06,
+      "loss": 0.0,
+      "step": 6140
+    },
+    {
+      "epoch": 90.44,
+      "grad_norm": 0.0019517322070896626,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.0,
+      "step": 6150
+    },
+    {
+      "epoch": 90.59,
+      "grad_norm": 0.0015430712373927236,
+      "learning_rate": 1.0545822102425878e-06,
+      "loss": 0.0,
+      "step": 6160
+    },
+    {
+      "epoch": 90.74,
+      "grad_norm": 0.0016941017238423228,
+      "learning_rate": 1.037735849056604e-06,
+      "loss": 0.0,
+      "step": 6170
+    },
+    {
+      "epoch": 90.88,
+      "grad_norm": 0.001742173102684319,
+      "learning_rate": 1.02088948787062e-06,
+      "loss": 0.0,
+      "step": 6180
+    },
+    {
+      "epoch": 91.03,
+      "grad_norm": 0.0018419664120301604,
+      "learning_rate": 1.0040431266846362e-06,
+      "loss": 0.0,
+      "step": 6190
+    },
+    {
+      "epoch": 91.18,
+      "grad_norm": 0.0017677543219178915,
+      "learning_rate": 9.871967654986524e-07,
+      "loss": 0.0,
+      "step": 6200
+    },
+    {
+      "epoch": 91.32,
+      "grad_norm": 0.0016228326130658388,
+      "learning_rate": 9.703504043126686e-07,
+      "loss": 0.0,
+      "step": 6210
+    },
+    {
+      "epoch": 91.47,
+      "grad_norm": 0.0014178988058120012,
+      "learning_rate": 9.535040431266847e-07,
+      "loss": 0.0,
+      "step": 6220
+    },
+    {
+      "epoch": 91.62,
+      "grad_norm": 0.0019479967886582017,
+      "learning_rate": 9.366576819407008e-07,
+      "loss": 0.0,
+      "step": 6230
+    },
+    {
+      "epoch": 91.76,
+      "grad_norm": 0.0013647832674905658,
+      "learning_rate": 9.19811320754717e-07,
+      "loss": 0.0,
+      "step": 6240
+    },
+    {
+      "epoch": 91.91,
+      "grad_norm": 0.001629057340323925,
+      "learning_rate": 9.029649595687333e-07,
+      "loss": 0.0,
+      "step": 6250
+    },
+    {
+      "epoch": 92.06,
+      "grad_norm": 0.0017082348931580782,
+      "learning_rate": 8.861185983827494e-07,
+      "loss": 0.0,
+      "step": 6260
+    },
+    {
+      "epoch": 92.21,
+      "grad_norm": 0.0017187732737511396,
+      "learning_rate": 8.692722371967656e-07,
+      "loss": 0.0,
+      "step": 6270
+    },
+    {
+      "epoch": 92.35,
+      "grad_norm": 0.0015693089226260781,
+      "learning_rate": 8.524258760107817e-07,
+      "loss": 0.0,
+      "step": 6280
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 0.0016504173399880528,
+      "learning_rate": 8.355795148247979e-07,
+      "loss": 0.0,
+      "step": 6290
+    },
+    {
+      "epoch": 92.65,
+      "grad_norm": 0.0014378475025296211,
+      "learning_rate": 8.18733153638814e-07,
+      "loss": 0.0,
+      "step": 6300
+    },
+    {
+      "epoch": 92.79,
+      "grad_norm": 0.0018091071397066116,
+      "learning_rate": 8.018867924528302e-07,
+      "loss": 0.0,
+      "step": 6310
+    },
+    {
+      "epoch": 92.94,
+      "grad_norm": 0.0016386961797252297,
+      "learning_rate": 7.850404312668463e-07,
+      "loss": 0.0,
+      "step": 6320
+    },
+    {
+      "epoch": 93.09,
+      "grad_norm": 0.0017092525959014893,
+      "learning_rate": 7.681940700808626e-07,
+      "loss": 0.0,
+      "step": 6330
+    },
+    {
+      "epoch": 93.24,
+      "grad_norm": 0.0015886749606579542,
+      "learning_rate": 7.513477088948788e-07,
+      "loss": 0.0,
+      "step": 6340
+    },
+    {
+      "epoch": 93.38,
+      "grad_norm": 0.0018031731015071273,
+      "learning_rate": 7.345013477088949e-07,
+      "loss": 0.0,
+      "step": 6350
+    },
+    {
+      "epoch": 93.53,
+      "grad_norm": 0.0015261101070791483,
+      "learning_rate": 7.176549865229111e-07,
+      "loss": 0.0,
+      "step": 6360
+    },
+    {
+      "epoch": 93.68,
+      "grad_norm": 0.001621345872990787,
+      "learning_rate": 7.008086253369272e-07,
+      "loss": 0.0,
+      "step": 6370
+    },
+    {
+      "epoch": 93.82,
+      "grad_norm": 0.0018881525611504912,
+      "learning_rate": 6.839622641509434e-07,
+      "loss": 0.0,
+      "step": 6380
+    },
+    {
+      "epoch": 93.97,
+      "grad_norm": 0.0014266808284446597,
+      "learning_rate": 6.671159029649596e-07,
+      "loss": 0.0,
+      "step": 6390
+    },
+    {
+      "epoch": 94.12,
+      "grad_norm": 0.0017150483326986432,
+      "learning_rate": 6.502695417789757e-07,
+      "loss": 0.0,
+      "step": 6400
+    },
+    {
+      "epoch": 94.26,
+      "grad_norm": 0.001493273302912712,
+      "learning_rate": 6.33423180592992e-07,
+      "loss": 0.0,
+      "step": 6410
+    },
+    {
+      "epoch": 94.41,
+      "grad_norm": 0.0016202511033043265,
+      "learning_rate": 6.165768194070082e-07,
+      "loss": 0.0,
+      "step": 6420
+    },
+    {
+      "epoch": 94.56,
+      "grad_norm": 0.001579651259817183,
+      "learning_rate": 5.997304582210243e-07,
+      "loss": 0.0,
+      "step": 6430
+    },
+    {
+      "epoch": 94.71,
+      "grad_norm": 0.0014305037911981344,
+      "learning_rate": 5.828840970350405e-07,
+      "loss": 0.0,
+      "step": 6440
+    },
+    {
+      "epoch": 94.85,
+      "grad_norm": 0.0015384092694148421,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 0.0,
+      "step": 6450
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 0.0018785832216963172,
+      "learning_rate": 5.491913746630729e-07,
+      "loss": 0.0,
+      "step": 6460
+    },
+    {
+      "epoch": 95.15,
+      "grad_norm": 0.001387153286486864,
+      "learning_rate": 5.32345013477089e-07,
+      "loss": 0.0,
+      "step": 6470
+    },
+    {
+      "epoch": 95.29,
+      "grad_norm": 0.0013170434394851327,
+      "learning_rate": 5.154986522911052e-07,
+      "loss": 0.0,
+      "step": 6480
+    },
+    {
+      "epoch": 95.44,
+      "grad_norm": 0.001672849990427494,
+      "learning_rate": 4.986522911051214e-07,
+      "loss": 0.0,
+      "step": 6490
+    },
+    {
+      "epoch": 95.59,
+      "grad_norm": 0.0014924348797649145,
+      "learning_rate": 4.818059299191375e-07,
+      "loss": 0.0,
+      "step": 6500
+    },
+    {
+      "epoch": 95.59,
+      "eval_loss": 0.5511065125465393,
+      "eval_runtime": 393.4846,
+      "eval_samples_per_second": 1.916,
+      "eval_steps_per_second": 0.02,
+      "eval_wer": 39.565217391304344,
+      "step": 6500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6784,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 100,
+  "save_steps": 500,
+  "total_flos": 1.87169276012544e+20,
+  "train_batch_size": 100,
+  "trial_name": null,
+  "trial_params": null
+}