End of training

README.md

@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-large-v3-ft-btb-ca-cy
 
-This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on an unknown dataset.
+This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, cymen-arfor/15awr train+dev main dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.4080
 - Wer: 0.2782

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 3.1938677738741617,
-    "eval_loss": 0.40369176864624023,
-    "eval_runtime": 1635.5488,
+    "epoch": 3.2776138970829236,
+    "eval_loss": 0.40801551938056946,
+    "eval_runtime": 1617.1708,
     "eval_samples": 3901,
-    "eval_samples_per_second": 2.385,
-    "eval_steps_per_second": 0.149,
-    "eval_wer": 0.27909784603496574,
+    "eval_samples_per_second": 2.412,
+    "eval_steps_per_second": 0.151,
+    "eval_wer": 0.2782285327924273,
     "total_flos": 5.435895365546803e+20,
-    "train_loss": 0.3537434986591339,
-    "train_runtime": 57744.1557,
-    "train_samples": 50095,
-    "train_samples_per_second": 2.771,
-    "train_steps_per_second": 0.087
+    "train_loss": 0.6995921184539795,
+    "train_runtime": 59796.6444,
+    "train_samples": 48815,
+    "train_samples_per_second": 2.676,
+    "train_steps_per_second": 0.084
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.1938677738741617,
-    "eval_loss": 0.40369176864624023,
-    "eval_runtime": 1635.5488,
+    "epoch": 3.2776138970829236,
+    "eval_loss": 0.40801551938056946,
+    "eval_runtime": 1617.1708,
     "eval_samples": 3901,
-    "eval_samples_per_second": 2.385,
-    "eval_steps_per_second": 0.149,
-    "eval_wer": 0.27909784603496574
+    "eval_samples_per_second": 2.412,
+    "eval_steps_per_second": 0.151,
+    "eval_wer": 0.2782285327924273
 }

runs/Nov04_13-33-16_daf84da58340/events.out.tfevents.1730790162.daf84da58340.2283.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9a6ddd8a7bc9bebd2b92930f1ca0b185e4a749155e6b441a95b3a2e871f2aec
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.1938677738741617,
+    "epoch": 3.2776138970829236,
     "total_flos": 5.435895365546803e+20,
-    "train_loss": 0.3537434986591339,
-    "train_runtime": 57744.1557,
-    "train_samples": 50095,
-    "train_samples_per_second": 2.771,
-    "train_steps_per_second": 0.087
+    "train_loss": 0.6995921184539795,
+    "train_runtime": 59796.6444,
+    "train_samples": 48815,
+    "train_samples_per_second": 2.676,
+    "train_steps_per_second": 0.084
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.1938677738741617,
+  "epoch": 3.2776138970829236,
   "eval_steps": 1000,
   "global_step": 5000,
   "is_hyper_param_search": false,
@@ -9,1458 +9,1458 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.015969338869370808,
-      "grad_norm": 9.112746238708496,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 1.6703,
+      "epoch": 0.01638806948541462,
+      "grad_norm": 16.889263153076172,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.3203,
       "step": 25
     },
     {
-      "epoch": 0.031938677738741615,
-      "grad_norm": 5.510941505432129,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.2963,
+      "epoch": 0.03277613897082924,
+      "grad_norm": 11.537278175354004,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.681,
       "step": 50
     },
     {
-      "epoch": 0.047908016608112426,
-      "grad_norm": 5.158988952636719,
-      "learning_rate": 1.5e-06,
-      "loss": 1.0164,
+      "epoch": 0.049164208456243856,
+      "grad_norm": 10.447038650512695,
+      "learning_rate": 1.48e-06,
+      "loss": 2.0602,
       "step": 75
     },
     {
-      "epoch": 0.06387735547748323,
-      "grad_norm": 5.4585957527160645,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.8123,
+      "epoch": 0.06555227794165848,
+      "grad_norm": 10.497332572937012,
+      "learning_rate": 1.98e-06,
+      "loss": 1.7035,
       "step": 100
     },
     {
-      "epoch": 0.07984669434685404,
-      "grad_norm": 5.480216979980469,
-      "learning_rate": 2.5e-06,
-      "loss": 0.7918,
+      "epoch": 0.0819403474270731,
+      "grad_norm": 10.367623329162598,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 1.6221,
       "step": 125
     },
     {
-      "epoch": 0.09581603321622485,
-      "grad_norm": 5.493738651275635,
-      "learning_rate": 3e-06,
-      "loss": 0.7452,
+      "epoch": 0.09832841691248771,
+      "grad_norm": 8.87679386138916,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 1.5737,
       "step": 150
     },
     {
-      "epoch": 0.11178537208559565,
-      "grad_norm": 4.575730800628662,
-      "learning_rate": 3.5e-06,
-      "loss": 0.7583,
+      "epoch": 0.11471648639790233,
+      "grad_norm": 10.882203102111816,
+      "learning_rate": 3.48e-06,
+      "loss": 1.4533,
       "step": 175
     },
     {
-      "epoch": 0.12775471095496646,
-      "grad_norm": 4.527775764465332,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.7108,
+      "epoch": 0.13110455588331696,
+      "grad_norm": 9.381525993347168,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 1.3633,
       "step": 200
     },
     {
-      "epoch": 0.14372404982433729,
-      "grad_norm": 4.807627201080322,
-      "learning_rate": 4.5e-06,
-      "loss": 0.7146,
+      "epoch": 0.14749262536873156,
+      "grad_norm": 9.403127670288086,
+      "learning_rate": 4.48e-06,
+      "loss": 1.3824,
       "step": 225
     },
     {
-      "epoch": 0.15969338869370808,
-      "grad_norm": 4.733044147491455,
-      "learning_rate": 5e-06,
-      "loss": 0.659,
+      "epoch": 0.1638806948541462,
+      "grad_norm": 9.816941261291504,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 1.3554,
       "step": 250
     },
     {
-      "epoch": 0.17566272756307888,
-      "grad_norm": 5.310563564300537,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 0.7,
+      "epoch": 0.1802687643395608,
+      "grad_norm": 9.448797225952148,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 1.311,
       "step": 275
     },
     {
-      "epoch": 0.1916320664324497,
-      "grad_norm": 4.584811687469482,
-      "learning_rate": 6e-06,
-      "loss": 0.6643,
+      "epoch": 0.19665683382497542,
+      "grad_norm": 9.280777931213379,
+      "learning_rate": 5.98e-06,
+      "loss": 1.2986,
       "step": 300
     },
     {
-      "epoch": 0.2076014053018205,
-      "grad_norm": 5.011808395385742,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.6696,
+      "epoch": 0.21304490331039003,
+      "grad_norm": 9.549225807189941,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 1.3045,
       "step": 325
     },
     {
-      "epoch": 0.2235707441711913,
-      "grad_norm": 4.633641719818115,
-      "learning_rate": 7e-06,
-      "loss": 0.6692,
+      "epoch": 0.22943297279580466,
+      "grad_norm": 11.352511405944824,
+      "learning_rate": 6.98e-06,
+      "loss": 1.2564,
       "step": 350
     },
     {
-      "epoch": 0.23954008304056212,
-      "grad_norm": 3.665330648422241,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.6285,
+      "epoch": 0.24582104228121926,
+      "grad_norm": 7.8544020652771,
+      "learning_rate": 7.48e-06,
+      "loss": 1.2104,
       "step": 375
     },
     {
-      "epoch": 0.2555094219099329,
-      "grad_norm": 5.0646185874938965,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.6313,
+      "epoch": 0.2622091117666339,
+      "grad_norm": 9.684502601623535,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 1.2606,
       "step": 400
     },
     {
-      "epoch": 0.2714787607793037,
-      "grad_norm": 3.9207818508148193,
-      "learning_rate": 8.5e-06,
-      "loss": 0.6045,
+      "epoch": 0.2785971812520485,
+      "grad_norm": 9.090190887451172,
+      "learning_rate": 8.48e-06,
+      "loss": 1.2554,
       "step": 425
     },
     {
-      "epoch": 0.28744809964867457,
-      "grad_norm": 4.316348552703857,
-      "learning_rate": 9e-06,
-      "loss": 0.6044,
+      "epoch": 0.2949852507374631,
+      "grad_norm": 8.529007911682129,
+      "learning_rate": 8.98e-06,
+      "loss": 1.2412,
       "step": 450
     },
     {
-      "epoch": 0.30341743851804537,
-      "grad_norm": 3.9778010845184326,
-      "learning_rate": 9.5e-06,
-      "loss": 0.6015,
+      "epoch": 0.3113733202228777,
+      "grad_norm": 8.554973602294922,
+      "learning_rate": 9.48e-06,
+      "loss": 1.2064,
       "step": 475
     },
     {
-      "epoch": 0.31938677738741617,
-      "grad_norm": 3.8584372997283936,
-      "learning_rate": 1e-05,
-      "loss": 0.5793,
+      "epoch": 0.3277613897082924,
+      "grad_norm": 7.774643421173096,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 1.2244,
       "step": 500
     },
     {
-      "epoch": 0.33535611625678696,
-      "grad_norm": 4.777307033538818,
-      "learning_rate": 9.944444444444445e-06,
-      "loss": 0.6289,
+      "epoch": 0.344149459193707,
+      "grad_norm": 10.1014404296875,
+      "learning_rate": 9.946666666666667e-06,
+      "loss": 1.1452,
       "step": 525
     },
     {
-      "epoch": 0.35132545512615776,
-      "grad_norm": 4.359010219573975,
-      "learning_rate": 9.88888888888889e-06,
-      "loss": 0.6047,
+      "epoch": 0.3605375286791216,
+      "grad_norm": 8.465851783752441,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 1.1455,
       "step": 550
     },
     {
-      "epoch": 0.3672947939955286,
-      "grad_norm": 4.476827621459961,
-      "learning_rate": 9.833333333333333e-06,
-      "loss": 0.5868,
+      "epoch": 0.3769255981645362,
+      "grad_norm": 6.562053203582764,
+      "learning_rate": 9.835555555555556e-06,
+      "loss": 1.1587,
       "step": 575
     },
     {
-      "epoch": 0.3832641328648994,
-      "grad_norm": 3.478997230529785,
-      "learning_rate": 9.777777777777779e-06,
-      "loss": 0.5877,
+      "epoch": 0.39331366764995085,
+      "grad_norm": 8.402182579040527,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 1.1842,
       "step": 600
     },
     {
-      "epoch": 0.3992334717342702,
-      "grad_norm": 4.191431522369385,
-      "learning_rate": 9.722222222222223e-06,
-      "loss": 0.547,
+      "epoch": 0.40970173713536545,
+      "grad_norm": 8.295367240905762,
+      "learning_rate": 9.724444444444445e-06,
+      "loss": 1.1326,
       "step": 625
     },
     {
-      "epoch": 0.415202810603641,
-      "grad_norm": 3.8496177196502686,
-      "learning_rate": 9.666666666666667e-06,
-      "loss": 0.5511,
+      "epoch": 0.42608980662078005,
+      "grad_norm": 8.351269721984863,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 1.1054,
       "step": 650
     },
     {
-      "epoch": 0.4311721494730118,
-      "grad_norm": 3.78070330619812,
-      "learning_rate": 9.611111111111112e-06,
-      "loss": 0.5503,
+      "epoch": 0.4424778761061947,
+      "grad_norm": 6.943824291229248,
+      "learning_rate": 9.613333333333335e-06,
+      "loss": 1.1294,
       "step": 675
     },
     {
-      "epoch": 0.4471414883423826,
-      "grad_norm": 3.223071813583374,
-      "learning_rate": 9.555555555555556e-06,
-      "loss": 0.5418,
+      "epoch": 0.4588659455916093,
+      "grad_norm": 8.862208366394043,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 1.0959,
       "step": 700
     },
     {
-      "epoch": 0.46311082721175345,
-      "grad_norm": 4.323662281036377,
-      "learning_rate": 9.5e-06,
-      "loss": 0.5362,
+      "epoch": 0.4752540150770239,
+      "grad_norm": 7.068411350250244,
+      "learning_rate": 9.502222222222223e-06,
+      "loss": 1.1214,
       "step": 725
     },
     {
-      "epoch": 0.47908016608112425,
-      "grad_norm": 3.7754769325256348,
-      "learning_rate": 9.444444444444445e-06,
-      "loss": 0.5478,
+      "epoch": 0.4916420845624385,
+      "grad_norm": 7.394863605499268,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 1.0502,
       "step": 750
     },
     {
-      "epoch": 0.49504950495049505,
-      "grad_norm": 4.361028671264648,
-      "learning_rate": 9.38888888888889e-06,
-      "loss": 0.5257,
+      "epoch": 0.5080301540478531,
+      "grad_norm": 7.007723331451416,
+      "learning_rate": 9.391111111111111e-06,
+      "loss": 0.9964,
       "step": 775
     },
     {
-      "epoch": 0.5110188438198658,
-      "grad_norm": 3.5216853618621826,
-      "learning_rate": 9.333333333333334e-06,
-      "loss": 0.5046,
+      "epoch": 0.5244182235332678,
+      "grad_norm": 8.071842193603516,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 1.0542,
       "step": 800
     },
     {
-      "epoch": 0.5269881826892366,
-      "grad_norm": 3.39201021194458,
-      "learning_rate": 9.277777777777778e-06,
-      "loss": 0.5209,
+      "epoch": 0.5408062930186824,
+      "grad_norm": 7.3528242111206055,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 1.0539,
       "step": 825
     },
     {
-      "epoch": 0.5429575215586074,
-      "grad_norm": 3.3704864978790283,
-      "learning_rate": 9.222222222222224e-06,
-      "loss": 0.5128,
+      "epoch": 0.557194362504097,
+      "grad_norm": 8.404199600219727,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 1.0574,
       "step": 850
     },
     {
-      "epoch": 0.5589268604279782,
-      "grad_norm": 3.4954028129577637,
-      "learning_rate": 9.166666666666666e-06,
-      "loss": 0.4954,
+      "epoch": 0.5735824319895116,
+      "grad_norm": 6.234879016876221,
+      "learning_rate": 9.168888888888889e-06,
+      "loss": 0.975,
       "step": 875
     },
     {
-      "epoch": 0.5748961992973491,
-      "grad_norm": 3.267204523086548,
-      "learning_rate": 9.111111111111112e-06,
-      "loss": 0.5204,
+      "epoch": 0.5899705014749262,
+      "grad_norm": 8.72309398651123,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 1.0259,
       "step": 900
     },
     {
-      "epoch": 0.5908655381667199,
-      "grad_norm": 3.802452802658081,
-      "learning_rate": 9.055555555555556e-06,
-      "loss": 0.4988,
+      "epoch": 0.6063585709603408,
+      "grad_norm": 6.828171730041504,
+      "learning_rate": 9.057777777777779e-06,
+      "loss": 1.0516,
       "step": 925
     },
     {
-      "epoch": 0.6068348770360907,
-      "grad_norm": 3.5180504322052,
-      "learning_rate": 9e-06,
-      "loss": 0.5068,
+      "epoch": 0.6227466404457555,
+      "grad_norm": 7.933355808258057,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 1.0096,
       "step": 950
     },
     {
-      "epoch": 0.6228042159054615,
-      "grad_norm": 3.5314347743988037,
-      "learning_rate": 8.944444444444446e-06,
-      "loss": 0.5039,
+      "epoch": 0.63913470993117,
+      "grad_norm": 9.511138916015625,
+      "learning_rate": 8.946666666666669e-06,
+      "loss": 1.0429,
       "step": 975
     },
     {
-      "epoch": 0.6387735547748323,
-      "grad_norm": 3.1311240196228027,
-      "learning_rate": 8.888888888888888e-06,
-      "loss": 0.4739,
+      "epoch": 0.6555227794165848,
+      "grad_norm": 6.3830108642578125,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.9864,
       "step": 1000
     },
     {
-      "epoch": 0.6387735547748323,
-      "eval_loss": 0.4840342402458191,
-      "eval_runtime": 1688.2428,
-      "eval_samples_per_second": 2.311,
-      "eval_steps_per_second": 0.145,
-      "eval_wer": 0.3517579445571332,
+      "epoch": 0.6555227794165848,
+      "eval_loss": 0.4853302240371704,
+      "eval_runtime": 1674.4328,
+      "eval_samples_per_second": 2.33,
+      "eval_steps_per_second": 0.146,
+      "eval_wer": 0.3735632183908046,
       "step": 1000
     },
     {
-      "epoch": 0.6547428936442031,
-      "grad_norm": 2.715566396713257,
-      "learning_rate": 8.833333333333334e-06,
-      "loss": 0.502,
+      "epoch": 0.6719108489019994,
+      "grad_norm": 7.367117404937744,
+      "learning_rate": 8.835555555555557e-06,
+      "loss": 1.0247,
       "step": 1025
     },
     {
-      "epoch": 0.6707122325135739,
-      "grad_norm": 3.692412853240967,
-      "learning_rate": 8.777777777777778e-06,
-      "loss": 0.4733,
+      "epoch": 0.688298918387414,
+      "grad_norm": 6.835860252380371,
+      "learning_rate": 8.78e-06,
+      "loss": 0.9806,
       "step": 1050
     },
     {
-      "epoch": 0.6866815713829447,
-      "grad_norm": 3.414935350418091,
-      "learning_rate": 8.722222222222224e-06,
-      "loss": 0.4945,
+      "epoch": 0.7046869878728286,
+      "grad_norm": 7.044719219207764,
+      "learning_rate": 8.724444444444445e-06,
+      "loss": 0.989,
       "step": 1075
     },
     {
-      "epoch": 0.7026509102523155,
-      "grad_norm": 3.058530569076538,
-      "learning_rate": 8.666666666666668e-06,
-      "loss": 0.5033,
+      "epoch": 0.7210750573582432,
+      "grad_norm": 7.264064311981201,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 1.0324,
       "step": 1100
     },
     {
-      "epoch": 0.7186202491216863,
-      "grad_norm": 3.8033292293548584,
-      "learning_rate": 8.611111111111112e-06,
-      "loss": 0.483,
+      "epoch": 0.7374631268436578,
+      "grad_norm": 6.116435527801514,
+      "learning_rate": 8.613333333333333e-06,
+      "loss": 0.9663,
       "step": 1125
     },
     {
-      "epoch": 0.7345895879910572,
-      "grad_norm": 3.238865852355957,
-      "learning_rate": 8.555555555555556e-06,
-      "loss": 0.4809,
+      "epoch": 0.7538511963290724,
+      "grad_norm": 6.0959672927856445,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.9516,
       "step": 1150
     },
     {
-      "epoch": 0.750558926860428,
-      "grad_norm": 2.665280342102051,
-      "learning_rate": 8.5e-06,
-      "loss": 0.4792,
+      "epoch": 0.7702392658144871,
+      "grad_norm": 7.6278486251831055,
+      "learning_rate": 8.502222222222223e-06,
+      "loss": 0.9008,
       "step": 1175
     },
     {
-      "epoch": 0.7665282657297988,
-      "grad_norm": 2.9003748893737793,
-      "learning_rate": 8.444444444444446e-06,
-      "loss": 0.4662,
+      "epoch": 0.7866273352999017,
+      "grad_norm": 6.617189884185791,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.9682,
       "step": 1200
     },
     {
-      "epoch": 0.7824976045991696,
-      "grad_norm": 2.7720658779144287,
-      "learning_rate": 8.38888888888889e-06,
-      "loss": 0.4532,
+      "epoch": 0.8030154047853163,
+      "grad_norm": 6.494309902191162,
+      "learning_rate": 8.391111111111112e-06,
+      "loss": 0.944,
       "step": 1225
     },
     {
-      "epoch": 0.7984669434685404,
-      "grad_norm": 3.0096704959869385,
-      "learning_rate": 8.333333333333334e-06,
-      "loss": 0.4719,
+      "epoch": 0.8194034742707309,
+      "grad_norm": 7.69552755355835,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.9454,
       "step": 1250
     },
     {
-      "epoch": 0.8144362823379112,
-      "grad_norm": 3.1118383407592773,
-      "learning_rate": 8.277777777777778e-06,
-      "loss": 0.4567,
+      "epoch": 0.8357915437561455,
+      "grad_norm": 5.991209506988525,
+      "learning_rate": 8.28e-06,
+      "loss": 0.9149,
       "step": 1275
     },
     {
-      "epoch": 0.830405621207282,
-      "grad_norm": 2.854698896408081,
-      "learning_rate": 8.222222222222222e-06,
-      "loss": 0.4729,
+      "epoch": 0.8521796132415601,
+      "grad_norm": 6.490001201629639,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.9106,
       "step": 1300
     },
     {
-      "epoch": 0.8463749600766528,
-      "grad_norm": 3.1824917793273926,
-      "learning_rate": 8.166666666666668e-06,
-      "loss": 0.5073,
+      "epoch": 0.8685676827269747,
+      "grad_norm": 6.849838733673096,
+      "learning_rate": 8.16888888888889e-06,
+      "loss": 0.8938,
       "step": 1325
     },
     {
-      "epoch": 0.8623442989460236,
-      "grad_norm": 3.554313898086548,
-      "learning_rate": 8.111111111111112e-06,
-      "loss": 0.4663,
+      "epoch": 0.8849557522123894,
+      "grad_norm": 5.19750452041626,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.9318,
       "step": 1350
     },
     {
-      "epoch": 0.8783136378153944,
-      "grad_norm": 4.958135604858398,
-      "learning_rate": 8.055555555555557e-06,
-      "loss": 0.4614,
+      "epoch": 0.901343821697804,
+      "grad_norm": 6.247749328613281,
+      "learning_rate": 8.057777777777778e-06,
+      "loss": 0.9398,
       "step": 1375
     },
     {
-      "epoch": 0.8942829766847652,
-      "grad_norm": 3.5128700733184814,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.4522,
+      "epoch": 0.9177318911832186,
+      "grad_norm": 9.035011291503906,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.9209,
       "step": 1400
     },
     {
-      "epoch": 0.9102523155541361,
-      "grad_norm": 4.640254020690918,
-      "learning_rate": 7.944444444444445e-06,
-      "loss": 0.4488,
+      "epoch": 0.9341199606686332,
+      "grad_norm": 6.385293960571289,
+      "learning_rate": 7.946666666666666e-06,
+      "loss": 0.9164,
       "step": 1425
     },
     {
-      "epoch": 0.9262216544235069,
-      "grad_norm": 2.928645610809326,
-      "learning_rate": 7.88888888888889e-06,
-      "loss": 0.4352,
+      "epoch": 0.9505080301540478,
+      "grad_norm": 6.2202348709106445,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.8646,
       "step": 1450
     },
     {
-      "epoch": 0.9421909932928777,
-      "grad_norm": 3.42668080329895,
-      "learning_rate": 7.833333333333333e-06,
-      "loss": 0.4648,
+      "epoch": 0.9668960996394624,
+      "grad_norm": 6.877343654632568,
+      "learning_rate": 7.835555555555556e-06,
+      "loss": 0.8866,
       "step": 1475
     },
     {
-      "epoch": 0.9581603321622485,
-      "grad_norm": 3.204360008239746,
-      "learning_rate": 7.77777777777778e-06,
-      "loss": 0.4318,
+      "epoch": 0.983284169124877,
+      "grad_norm": 7.396777153015137,
+      "learning_rate": 7.78e-06,
+      "loss": 0.9055,
       "step": 1500
     },
     {
-      "epoch": 0.9741296710316193,
-      "grad_norm": 2.9323625564575195,
-      "learning_rate": 7.722222222222223e-06,
-      "loss": 0.4464,
+      "epoch": 0.9996722386102918,
+      "grad_norm": 6.3318400382995605,
+      "learning_rate": 7.724444444444446e-06,
+      "loss": 0.8476,
       "step": 1525
     },
     {
-      "epoch": 0.9900990099009901,
-      "grad_norm": 3.103804349899292,
-      "learning_rate": 7.666666666666667e-06,
-      "loss": 0.4599,
+      "epoch": 1.0160603080957062,
+      "grad_norm": 5.315908432006836,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.666,
       "step": 1550
     },
     {
-      "epoch": 1.006068348770361,
-      "grad_norm": 2.95595121383667,
-      "learning_rate": 7.611111111111111e-06,
-      "loss": 0.4017,
+      "epoch": 1.0324483775811208,
+      "grad_norm": 4.697606086730957,
+      "learning_rate": 7.613333333333334e-06,
+      "loss": 0.6379,
       "step": 1575
     },
     {
-      "epoch": 1.0220376876397317,
-      "grad_norm": 2.5740716457366943,
-      "learning_rate": 7.555555555555556e-06,
-      "loss": 0.3208,
+      "epoch": 1.0488364470665354,
+      "grad_norm": 4.47541618347168,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.6537,
       "step": 1600
     },
     {
-      "epoch": 1.0380070265091026,
-      "grad_norm": 3.569032907485962,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.3206,
+      "epoch": 1.0652245165519503,
+      "grad_norm": 6.277323246002197,
+      "learning_rate": 7.502222222222223e-06,
+      "loss": 0.67,
       "step": 1625
     },
     {
-      "epoch": 1.0539763653784733,
-      "grad_norm": 3.01520037651062,
-      "learning_rate": 7.444444444444445e-06,
-      "loss": 0.325,
+      "epoch": 1.0816125860373649,
+      "grad_norm": 5.60599422454834,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.6839,
       "step": 1650
     },
     {
-      "epoch": 1.0699457042478442,
-      "grad_norm": 2.9796643257141113,
-      "learning_rate": 7.38888888888889e-06,
-      "loss": 0.3197,
+      "epoch": 1.0980006555227795,
+      "grad_norm": 5.50879430770874,
+      "learning_rate": 7.3911111111111125e-06,
+      "loss": 0.6778,
       "step": 1675
     },
     {
-      "epoch": 1.0859150431172149,
-      "grad_norm": 2.679849624633789,
-      "learning_rate": 7.333333333333333e-06,
-      "loss": 0.3305,
+      "epoch": 1.114388725008194,
+      "grad_norm": 7.369528293609619,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.681,
       "step": 1700
     },
     {
-      "epoch": 1.1018843819865858,
-      "grad_norm": 2.760922908782959,
-      "learning_rate": 7.277777777777778e-06,
-      "loss": 0.3333,
+      "epoch": 1.1307767944936087,
+      "grad_norm": 5.978742599487305,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 0.656,
       "step": 1725
     },
     {
-      "epoch": 1.1178537208559565,
-      "grad_norm": 2.7702419757843018,
-      "learning_rate": 7.222222222222223e-06,
-      "loss": 0.3452,
+      "epoch": 1.1471648639790233,
+      "grad_norm": 6.425286769866943,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.6781,
       "step": 1750
     },
     {
-      "epoch": 1.1338230597253274,
-      "grad_norm": 3.0105628967285156,
-      "learning_rate": 7.166666666666667e-06,
-      "loss": 0.3318,
+      "epoch": 1.1635529334644379,
+      "grad_norm": 5.3851165771484375,
+      "learning_rate": 7.1688888888888895e-06,
+      "loss": 0.6764,
       "step": 1775
     },
     {
-      "epoch": 1.1497923985946983,
-      "grad_norm": 3.0848121643066406,
-      "learning_rate": 7.111111111111112e-06,
-      "loss": 0.314,
+      "epoch": 1.1799410029498525,
+      "grad_norm": 5.43864631652832,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.6667,
       "step": 1800
     },
     {
-      "epoch": 1.165761737464069,
-      "grad_norm": 3.0407936573028564,
-      "learning_rate": 7.055555555555557e-06,
-      "loss": 0.339,
+      "epoch": 1.196329072435267,
+      "grad_norm": 5.792993545532227,
+      "learning_rate": 7.057777777777778e-06,
+      "loss": 0.6428,
       "step": 1825
     },
     {
-      "epoch": 1.1817310763334399,
-      "grad_norm": 2.692070722579956,
-      "learning_rate": 7e-06,
-      "loss": 0.3301,
+      "epoch": 1.2127171419206817,
+      "grad_norm": 6.183994293212891,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.6548,
       "step": 1850
     },
     {
-      "epoch": 1.1977004152028106,
-      "grad_norm": 2.7906270027160645,
-      "learning_rate": 6.944444444444445e-06,
-      "loss": 0.3243,
+      "epoch": 1.2291052114060963,
+      "grad_norm": 5.863783359527588,
+      "learning_rate": 6.946666666666667e-06,
+      "loss": 0.6681,
       "step": 1875
     },
     {
-      "epoch": 1.2136697540721815,
-      "grad_norm": 3.005059003829956,
-      "learning_rate": 6.88888888888889e-06,
-      "loss": 0.3348,
+      "epoch": 1.245493280891511,
+      "grad_norm": 6.144175052642822,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.6617,
       "step": 1900
     },
     {
-      "epoch": 1.2296390929415522,
-      "grad_norm": 2.512002944946289,
-      "learning_rate": 6.833333333333334e-06,
-      "loss": 0.3282,
+      "epoch": 1.2618813503769255,
+      "grad_norm": 5.673381805419922,
+      "learning_rate": 6.835555555555556e-06,
+      "loss": 0.6943,
       "step": 1925
     },
     {
-      "epoch": 1.245608431810923,
-      "grad_norm": 2.2834126949310303,
-      "learning_rate": 6.777777777777779e-06,
-      "loss": 0.3184,
+      "epoch": 1.2782694198623403,
+      "grad_norm": 5.691863536834717,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.6544,
       "step": 1950
     },
     {
-      "epoch": 1.2615777706802938,
-      "grad_norm": 2.6210882663726807,
-      "learning_rate": 6.7222222222222235e-06,
-      "loss": 0.3223,
+      "epoch": 1.2946574893477547,
+      "grad_norm": 5.3002028465271,
+      "learning_rate": 6.724444444444444e-06,
+      "loss": 0.6818,
       "step": 1975
     },
     {
-      "epoch": 1.2775471095496647,
-      "grad_norm": 2.5650439262390137,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.3335,
+      "epoch": 1.3110455588331695,
+      "grad_norm": 5.667751789093018,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.6533,
       "step": 2000
     },
     {
-      "epoch": 1.2775471095496647,
-      "eval_loss": 0.41682446002960205,
-      "eval_runtime": 1686.343,
-      "eval_samples_per_second": 2.313,
-      "eval_steps_per_second": 0.145,
-      "eval_wer": 0.31297691490389257,
+      "epoch": 1.3110455588331695,
+      "eval_loss": 0.4186985492706299,
+      "eval_runtime": 1629.6606,
+      "eval_samples_per_second": 2.394,
+      "eval_steps_per_second": 0.15,
+      "eval_wer": 0.3087752342316237,
       "step": 2000
     },
     {
-      "epoch": 1.2935164484190356,
-      "grad_norm": 3.5779247283935547,
-      "learning_rate": 6.6111111111111115e-06,
-      "loss": 0.3204,
+      "epoch": 1.3274336283185841,
+      "grad_norm": 5.364522933959961,
+      "learning_rate": 6.613333333333334e-06,
+      "loss": 0.6681,
       "step": 2025
     },
     {
-      "epoch": 1.3094857872884063,
-      "grad_norm": 2.3297035694122314,
-      "learning_rate": 6.555555555555556e-06,
-      "loss": 0.3237,
+      "epoch": 1.3438216978039987,
+      "grad_norm": 5.615537166595459,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.63,
       "step": 2050
     },
     {
-      "epoch": 1.325455126157777,
-      "grad_norm": 2.9575600624084473,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.3333,
+      "epoch": 1.3602097672894133,
+      "grad_norm": 5.930607318878174,
+      "learning_rate": 6.502222222222223e-06,
+      "loss": 0.6377,
       "step": 2075
     },
     {
-      "epoch": 1.3414244650271478,
-      "grad_norm": 2.6735305786132812,
-      "learning_rate": 6.444444444444445e-06,
-      "loss": 0.3199,
+      "epoch": 1.376597836774828,
+      "grad_norm": 8.01352310180664,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.6518,
       "step": 2100
     },
     {
-      "epoch": 1.3573938038965188,
-      "grad_norm": 3.5485923290252686,
-      "learning_rate": 6.3888888888888885e-06,
-      "loss": 0.3148,
+      "epoch": 1.3929859062602425,
+      "grad_norm": 5.568446159362793,
+      "learning_rate": 6.391111111111111e-06,
+      "loss": 0.6521,
       "step": 2125
     },
     {
-      "epoch": 1.3733631427658894,
-      "grad_norm": 2.704557180404663,
-      "learning_rate": 6.333333333333333e-06,
-      "loss": 0.3268,
+      "epoch": 1.4093739757456571,
+      "grad_norm": 5.956620693206787,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.6457,
       "step": 2150
     },
     {
-      "epoch": 1.3893324816352604,
-      "grad_norm": 2.417282819747925,
-      "learning_rate": 6.277777777777778e-06,
-      "loss": 0.3286,
+      "epoch": 1.4257620452310718,
+      "grad_norm": 5.39890193939209,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.6438,
       "step": 2175
     },
     {
-      "epoch": 1.405301820504631,
-      "grad_norm": 2.5613949298858643,
-      "learning_rate": 6.222222222222223e-06,
-      "loss": 0.3182,
+      "epoch": 1.4421501147164864,
+      "grad_norm": 5.913619518280029,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.6252,
       "step": 2200
     },
     {
-      "epoch": 1.421271159374002,
-      "grad_norm": 3.0761849880218506,
-      "learning_rate": 6.166666666666667e-06,
-      "loss": 0.329,
+      "epoch": 1.458538184201901,
+      "grad_norm": 4.932702541351318,
+      "learning_rate": 6.16888888888889e-06,
+      "loss": 0.6254,
       "step": 2225
     },
     {
-      "epoch": 1.4372404982433729,
-      "grad_norm": 2.764951229095459,
-      "learning_rate": 6.111111111111112e-06,
-      "loss": 0.3157,
+      "epoch": 1.4749262536873156,
+      "grad_norm": 5.434247970581055,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.6087,
       "step": 2250
     },
     {
-      "epoch": 1.4532098371127435,
-      "grad_norm": 2.909294605255127,
-      "learning_rate": 6.055555555555555e-06,
-      "loss": 0.3291,
+      "epoch": 1.4913143231727304,
+      "grad_norm": 6.103094577789307,
+      "learning_rate": 6.057777777777778e-06,
+      "loss": 0.6218,
       "step": 2275
     },
     {
-      "epoch": 1.4691791759821142,
-      "grad_norm": 3.374781370162964,
-      "learning_rate": 6e-06,
-      "loss": 0.3325,
+      "epoch": 1.5077023926581448,
+      "grad_norm": 5.433951377868652,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.6463,
       "step": 2300
     },
     {
-      "epoch": 1.4851485148514851,
-      "grad_norm": 2.8524813652038574,
-      "learning_rate": 5.944444444444445e-06,
-      "loss": 0.3182,
+      "epoch": 1.5240904621435596,
+      "grad_norm": 6.556436538696289,
+      "learning_rate": 5.946666666666668e-06,
+      "loss": 0.6431,
       "step": 2325
     },
     {
-      "epoch": 1.501117853720856,
-      "grad_norm": 3.066960573196411,
-      "learning_rate": 5.88888888888889e-06,
-      "loss": 0.3125,
+      "epoch": 1.540478531628974,
+      "grad_norm": 5.875002861022949,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.6461,
       "step": 2350
     },
     {
-      "epoch": 1.5170871925902267,
-      "grad_norm": 2.604304790496826,
-      "learning_rate": 5.833333333333334e-06,
-      "loss": 0.335,
+      "epoch": 1.5568666011143888,
+      "grad_norm": 6.182672500610352,
+      "learning_rate": 5.8355555555555565e-06,
+      "loss": 0.6393,
       "step": 2375
     },
     {
-      "epoch": 1.5330565314595974,
-      "grad_norm": 2.784276008605957,
-      "learning_rate": 5.777777777777778e-06,
-      "loss": 0.3318,
+      "epoch": 1.5732546705998034,
+      "grad_norm": 5.487701416015625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.6271,
       "step": 2400
     },
     {
-      "epoch": 1.5490258703289683,
-      "grad_norm": 3.203049898147583,
-      "learning_rate": 5.722222222222222e-06,
-      "loss": 0.3208,
+      "epoch": 1.589642740085218,
+      "grad_norm": 5.09088134765625,
+      "learning_rate": 5.724444444444445e-06,
+      "loss": 0.6203,
       "step": 2425
     },
     {
-      "epoch": 1.5649952091983392,
-      "grad_norm": 2.73185658454895,
-      "learning_rate": 5.666666666666667e-06,
-      "loss": 0.2981,
+      "epoch": 1.6060308095706326,
+      "grad_norm": 6.269595623016357,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.6279,
       "step": 2450
     },
     {
-      "epoch": 1.5809645480677101,
-      "grad_norm": 2.6829750537872314,
-      "learning_rate": 5.611111111111112e-06,
-      "loss": 0.3191,
+      "epoch": 1.6224188790560472,
+      "grad_norm": 6.465939998626709,
+      "learning_rate": 5.613333333333334e-06,
+      "loss": 0.7131,
       "step": 2475
     },
     {
-      "epoch": 1.5969338869370808,
-      "grad_norm": 3.1976850032806396,
-      "learning_rate": 5.555555555555557e-06,
-      "loss": 0.3335,
+      "epoch": 1.6388069485414618,
+      "grad_norm": 5.981501579284668,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.6274,
       "step": 2500
     },
     {
-      "epoch": 1.6129032258064515,
-      "grad_norm": 3.003782033920288,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 0.3013,
+      "epoch": 1.6551950180268764,
+      "grad_norm": 5.8698039054870605,
+      "learning_rate": 5.5022222222222224e-06,
+      "loss": 0.6282,
       "step": 2525
     },
     {
-      "epoch": 1.6288725646758224,
-      "grad_norm": 2.897247076034546,
-      "learning_rate": 5.444444444444445e-06,
-      "loss": 0.3123,
+      "epoch": 1.671583087512291,
+      "grad_norm": 5.132990837097168,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.6393,
       "step": 2550
     },
     {
-      "epoch": 1.6448419035451933,
-      "grad_norm": 2.996609687805176,
-      "learning_rate": 5.388888888888889e-06,
-      "loss": 0.3182,
+      "epoch": 1.6879711569977056,
+      "grad_norm": 5.995158672332764,
+      "learning_rate": 5.391111111111111e-06,
+      "loss": 0.6261,
       "step": 2575
     },
     {
-      "epoch": 1.660811242414564,
-      "grad_norm": 2.700706720352173,
-      "learning_rate": 5.333333333333334e-06,
-      "loss": 0.312,
+      "epoch": 1.7043592264831204,
+      "grad_norm": 6.151627540588379,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.6206,
       "step": 2600
     },
     {
-      "epoch": 1.6767805812839347,
-      "grad_norm": 2.901888370513916,
-      "learning_rate": 5.2777777777777785e-06,
-      "loss": 0.3088,
+      "epoch": 1.7207472959685348,
+      "grad_norm": 5.593701362609863,
+      "learning_rate": 5.28e-06,
+      "loss": 0.6193,
       "step": 2625
     },
     {
-      "epoch": 1.6927499201533056,
-      "grad_norm": 2.958951711654663,
-      "learning_rate": 5.2222222222222226e-06,
-      "loss": 0.3283,
+      "epoch": 1.7371353654539496,
+      "grad_norm": 5.100614547729492,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.6222,
       "step": 2650
     },
     {
-      "epoch": 1.7087192590226765,
-      "grad_norm": 3.610682964324951,
-      "learning_rate": 5.1666666666666675e-06,
-      "loss": 0.3332,
+      "epoch": 1.753523434939364,
+      "grad_norm": 5.8003411293029785,
+      "learning_rate": 5.168888888888889e-06,
+      "loss": 0.6356,
       "step": 2675
     },
     {
-      "epoch": 1.7246885978920474,
-      "grad_norm": 2.292649745941162,
-      "learning_rate": 5.1111111111111115e-06,
-      "loss": 0.2918,
+      "epoch": 1.7699115044247788,
+      "grad_norm": 5.680091857910156,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.6037,
       "step": 2700
     },
     {
-      "epoch": 1.7406579367614181,
-      "grad_norm": 2.41726016998291,
-      "learning_rate": 5.0555555555555555e-06,
-      "loss": 0.3215,
+      "epoch": 1.7862995739101932,
+      "grad_norm": 5.687478542327881,
+      "learning_rate": 5.057777777777778e-06,
+      "loss": 0.6538,
       "step": 2725
     },
     {
-      "epoch": 1.7566272756307888,
-      "grad_norm": 2.4622721672058105,
-      "learning_rate": 5e-06,
-      "loss": 0.3025,
+      "epoch": 1.802687643395608,
+      "grad_norm": 5.072441577911377,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.6083,
       "step": 2750
     },
     {
-      "epoch": 1.7725966145001597,
-      "grad_norm": 2.802022695541382,
-      "learning_rate": 4.944444444444445e-06,
-      "loss": 0.3008,
+      "epoch": 1.8190757128810227,
+      "grad_norm": 7.178542613983154,
+      "learning_rate": 4.946666666666667e-06,
+      "loss": 0.6398,
       "step": 2775
     },
     {
-      "epoch": 1.7885659533695306,
-      "grad_norm": 2.8401265144348145,
-      "learning_rate": 4.888888888888889e-06,
-      "loss": 0.3065,
+      "epoch": 1.8354637823664373,
+      "grad_norm": 6.016773223876953,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.635,
       "step": 2800
     },
     {
-      "epoch": 1.8045352922389013,
-      "grad_norm": 2.6890196800231934,
-      "learning_rate": 4.833333333333333e-06,
-      "loss": 0.3052,
+      "epoch": 1.8518518518518519,
+      "grad_norm": 5.279956817626953,
+      "learning_rate": 4.835555555555556e-06,
+      "loss": 0.6092,
       "step": 2825
     },
     {
-      "epoch": 1.820504631108272,
-      "grad_norm": 3.257754325866699,
-      "learning_rate": 4.777777777777778e-06,
-      "loss": 0.3243,
+      "epoch": 1.8682399213372665,
+      "grad_norm": 5.370604515075684,
+      "learning_rate": 4.78e-06,
+      "loss": 0.6303,
       "step": 2850
     },
     {
-      "epoch": 1.836473969977643,
-      "grad_norm": 2.595125675201416,
-      "learning_rate": 4.722222222222222e-06,
-      "loss": 0.316,
+      "epoch": 1.884627990822681,
+      "grad_norm": 4.235498428344727,
+      "learning_rate": 4.724444444444445e-06,
+      "loss": 0.6349,
       "step": 2875
     },
     {
-      "epoch": 1.8524433088470138,
-      "grad_norm": 2.713214159011841,
-      "learning_rate": 4.666666666666667e-06,
-      "loss": 0.3162,
+      "epoch": 1.9010160603080957,
+      "grad_norm": 4.0051374435424805,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.621,
       "step": 2900
     },
     {
-      "epoch": 1.8684126477163845,
-      "grad_norm": 2.4735445976257324,
-      "learning_rate": 4.611111111111112e-06,
-      "loss": 0.2955,
+      "epoch": 1.9174041297935103,
+      "grad_norm": 4.7797980308532715,
+      "learning_rate": 4.613333333333334e-06,
+      "loss": 0.6024,
       "step": 2925
     },
     {
-      "epoch": 1.8843819865857554,
-      "grad_norm": 3.026921033859253,
+      "epoch": 1.9337921992789249,
+      "grad_norm": 4.54160213470459,
       "learning_rate": 4.557777777777778e-06,
-      "loss": 0.3058,
+      "loss": 0.6105,
       "step": 2950
     },
     {
-      "epoch": 1.900351325455126,
-      "grad_norm": 2.5658116340637207,
+      "epoch": 1.9501802687643397,
+      "grad_norm": 5.6936259269714355,
       "learning_rate": 4.502222222222223e-06,
-      "loss": 0.3107,
+      "loss": 0.5948,
       "step": 2975
     },
     {
-      "epoch": 1.916320664324497,
-      "grad_norm": 2.6040449142456055,
+      "epoch": 1.966568338249754,
+      "grad_norm": 5.465245246887207,
       "learning_rate": 4.446666666666667e-06,
-      "loss": 0.3281,
+      "loss": 0.5981,
       "step": 3000
     },
     {
-      "epoch": 1.916320664324497,
-      "eval_loss": 0.3877977430820465,
-      "eval_runtime": 1665.3573,
-      "eval_samples_per_second": 2.342,
-      "eval_steps_per_second": 0.147,
-      "eval_wer": 0.2965324060658746,
+      "epoch": 1.966568338249754,
+      "eval_loss": 0.3845435380935669,
+      "eval_runtime": 1616.7761,
+      "eval_samples_per_second": 2.413,
+      "eval_steps_per_second": 0.151,
+      "eval_wer": 0.2868009272674587,
       "step": 3000
     },
     {
-      "epoch": 1.932290003193868,
-      "grad_norm": 2.7531864643096924,
+      "epoch": 1.982956407735169,
+      "grad_norm": 4.7660112380981445,
       "learning_rate": 4.391111111111112e-06,
-      "loss": 0.3129,
+      "loss": 0.6161,
       "step": 3025
     },
     {
-      "epoch": 1.9482593420632386,
-      "grad_norm": 2.1554644107818604,
+      "epoch": 1.9993444772205833,
+      "grad_norm": 5.041405200958252,
       "learning_rate": 4.3355555555555565e-06,
-      "loss": 0.2996,
+      "loss": 0.6114,
       "step": 3050
     },
     {
-      "epoch": 1.9642286809326093,
-      "grad_norm": 2.7987213134765625,
+      "epoch": 2.015732546705998,
+      "grad_norm": 4.265076160430908,
       "learning_rate": 4.2800000000000005e-06,
-      "loss": 0.3048,
+      "loss": 0.4373,
       "step": 3075
     },
     {
-      "epoch": 1.9801980198019802,
-      "grad_norm": 2.697014093399048,
+      "epoch": 2.0321206161914125,
+      "grad_norm": 5.111143589019775,
       "learning_rate": 4.2244444444444446e-06,
-      "loss": 0.3101,
+      "loss": 0.3853,
       "step": 3100
     },
     {
-      "epoch": 1.996167358671351,
-      "grad_norm": 2.2209935188293457,
+      "epoch": 2.0485086856768273,
+      "grad_norm": 4.114192962646484,
       "learning_rate": 4.168888888888889e-06,
-      "loss": 0.2935,
+      "loss": 0.3878,
       "step": 3125
     },
     {
-      "epoch": 2.012136697540722,
-      "grad_norm": 2.877319574356079,
+      "epoch": 2.0648967551622417,
+      "grad_norm": 5.525832176208496,
       "learning_rate": 4.1133333333333335e-06,
-      "loss": 0.2273,
+      "loss": 0.4061,
       "step": 3150
     },
     {
-      "epoch": 2.0281060364100925,
-      "grad_norm": 2.122941732406616,
+      "epoch": 2.0812848246476565,
+      "grad_norm": 5.436038494110107,
       "learning_rate": 4.057777777777778e-06,
-      "loss": 0.1945,
+      "loss": 0.395,
       "step": 3175
     },
     {
-      "epoch": 2.0440753752794634,
-      "grad_norm": 2.2523446083068848,
+      "epoch": 2.097672894133071,
+      "grad_norm": 4.541766166687012,
       "learning_rate": 4.002222222222222e-06,
-      "loss": 0.198,
+      "loss": 0.4178,
       "step": 3200
     },
     {
-      "epoch": 2.0600447141488343,
-      "grad_norm": 2.1842360496520996,
+      "epoch": 2.1140609636184857,
+      "grad_norm": 5.008460998535156,
       "learning_rate": 3.946666666666667e-06,
-      "loss": 0.1964,
+      "loss": 0.3989,
       "step": 3225
     },
     {
-      "epoch": 2.076014053018205,
-      "grad_norm": 2.5835633277893066,
+      "epoch": 2.1304490331039005,
+      "grad_norm": 5.069697380065918,
       "learning_rate": 3.891111111111111e-06,
-      "loss": 0.2093,
+      "loss": 0.4104,
       "step": 3250
     },
     {
-      "epoch": 2.0919833918875756,
-      "grad_norm": 2.534207344055176,
+      "epoch": 2.146837102589315,
+      "grad_norm": 5.150877952575684,
       "learning_rate": 3.835555555555555e-06,
-      "loss": 0.193,
+      "loss": 0.4088,
       "step": 3275
     },
     {
-      "epoch": 2.1079527307569466,
-      "grad_norm": 3.2382678985595703,
+      "epoch": 2.1632251720747298,
+      "grad_norm": 4.097008228302002,
       "learning_rate": 3.7800000000000002e-06,
-      "loss": 0.2066,
+      "loss": 0.3898,
       "step": 3300
     },
     {
-      "epoch": 2.1239220696263175,
-      "grad_norm": 1.761076807975769,
+      "epoch": 2.179613241560144,
+      "grad_norm": 5.378806114196777,
       "learning_rate": 3.724444444444445e-06,
-      "loss": 0.2032,
+      "loss": 0.404,
       "step": 3325
     },
     {
-      "epoch": 2.1398914084956884,
-      "grad_norm": 2.6212522983551025,
+      "epoch": 2.196001311045559,
+      "grad_norm": 4.335798263549805,
       "learning_rate": 3.668888888888889e-06,
-      "loss": 0.2018,
+      "loss": 0.3919,
       "step": 3350
     },
     {
-      "epoch": 2.1558607473650593,
-      "grad_norm": 2.240907669067383,
+      "epoch": 2.2123893805309733,
+      "grad_norm": 5.22952938079834,
       "learning_rate": 3.6133333333333336e-06,
-      "loss": 0.203,
+      "loss": 0.3805,
       "step": 3375
     },
     {
-      "epoch": 2.1718300862344297,
-      "grad_norm": 2.5074503421783447,
+      "epoch": 2.228777450016388,
+      "grad_norm": 4.839203357696533,
       "learning_rate": 3.5577777777777785e-06,
-      "loss": 0.2029,
+      "loss": 0.3778,
       "step": 3400
     },
     {
-      "epoch": 2.1877994251038007,
-      "grad_norm": 2.3992538452148438,
+      "epoch": 2.2451655195018025,
+      "grad_norm": 5.095648765563965,
       "learning_rate": 3.5022222222222225e-06,
-      "loss": 0.1975,
+      "loss": 0.4234,
       "step": 3425
     },
     {
-      "epoch": 2.2037687639731716,
-      "grad_norm": 2.3788444995880127,
+      "epoch": 2.2615535889872174,
+      "grad_norm": 5.199317455291748,
       "learning_rate": 3.446666666666667e-06,
-      "loss": 0.2004,
+      "loss": 0.394,
       "step": 3450
     },
     {
-      "epoch": 2.2197381028425425,
-      "grad_norm": 2.7057225704193115,
+      "epoch": 2.2779416584726317,
+      "grad_norm": 4.367405891418457,
       "learning_rate": 3.391111111111111e-06,
-      "loss": 0.1956,
+      "loss": 0.4097,
       "step": 3475
     },
     {
-      "epoch": 2.235707441711913,
-      "grad_norm": 2.399021863937378,
+      "epoch": 2.2943297279580466,
+      "grad_norm": 4.368666648864746,
       "learning_rate": 3.335555555555556e-06,
-      "loss": 0.2033,
+      "loss": 0.3983,
       "step": 3500
     },
     {
-      "epoch": 2.251676780581284,
-      "grad_norm": 2.3665523529052734,
+      "epoch": 2.310717797443461,
+      "grad_norm": 5.194695472717285,
       "learning_rate": 3.2800000000000004e-06,
-      "loss": 0.203,
+      "loss": 0.391,
       "step": 3525
     },
     {
-      "epoch": 2.2676461194506548,
-      "grad_norm": 2.4822895526885986,
+      "epoch": 2.3271058669288758,
+      "grad_norm": 5.09294319152832,
       "learning_rate": 3.2244444444444444e-06,
-      "loss": 0.1991,
+      "loss": 0.3773,
       "step": 3550
     },
     {
-      "epoch": 2.2836154583200257,
-      "grad_norm": 2.2787749767303467,
+      "epoch": 2.3434939364142906,
+      "grad_norm": 4.393428325653076,
       "learning_rate": 3.1688888888888893e-06,
-      "loss": 0.2024,
+      "loss": 0.3953,
       "step": 3575
     },
     {
-      "epoch": 2.2995847971893966,
-      "grad_norm": 2.3927159309387207,
+      "epoch": 2.359882005899705,
+      "grad_norm": 4.545704364776611,
       "learning_rate": 3.1133333333333337e-06,
-      "loss": 0.1928,
+      "loss": 0.4155,
       "step": 3600
     },
     {
-      "epoch": 2.315554136058767,
-      "grad_norm": 2.311298370361328,
+      "epoch": 2.37627007538512,
+      "grad_norm": 5.0112786293029785,
       "learning_rate": 3.0577777777777778e-06,
-      "loss": 0.2231,
+      "loss": 0.4746,
       "step": 3625
     },
     {
-      "epoch": 2.331523474928138,
-      "grad_norm": 2.4807193279266357,
+      "epoch": 2.392658144870534,
+      "grad_norm": 4.352410316467285,
       "learning_rate": 3.0022222222222227e-06,
-      "loss": 0.1938,
+      "loss": 0.4072,
       "step": 3650
     },
     {
-      "epoch": 2.347492813797509,
-      "grad_norm": 2.698840856552124,
+      "epoch": 2.409046214355949,
+      "grad_norm": 5.26077938079834,
       "learning_rate": 2.946666666666667e-06,
-      "loss": 0.1954,
+      "loss": 0.3907,
       "step": 3675
     },
     {
-      "epoch": 2.3634621526668798,
-      "grad_norm": 2.193194627761841,
+      "epoch": 2.4254342838413634,
+      "grad_norm": 5.140880107879639,
       "learning_rate": 2.891111111111111e-06,
-      "loss": 0.2013,
+      "loss": 0.4167,
       "step": 3700
     },
     {
-      "epoch": 2.37943149153625,
-      "grad_norm": 2.5082945823669434,
+      "epoch": 2.441822353326778,
+      "grad_norm": 4.907773494720459,
       "learning_rate": 2.835555555555556e-06,
-      "loss": 0.1963,
+      "loss": 0.3951,
       "step": 3725
     },
     {
-      "epoch": 2.395400830405621,
-      "grad_norm": 2.7638180255889893,
+      "epoch": 2.4582104228121926,
+      "grad_norm": 3.897996425628662,
       "learning_rate": 2.7800000000000005e-06,
-      "loss": 0.1919,
+      "loss": 0.3899,
       "step": 3750
     },
     {
-      "epoch": 2.411370169274992,
-      "grad_norm": 2.6804022789001465,
+      "epoch": 2.4745984922976074,
+      "grad_norm": 5.431221008300781,
       "learning_rate": 2.7244444444444445e-06,
-      "loss": 0.2008,
+      "loss": 0.4015,
       "step": 3775
     },
     {
-      "epoch": 2.427339508144363,
-      "grad_norm": 2.131471872329712,
+      "epoch": 2.490986561783022,
+      "grad_norm": 6.41071081161499,
       "learning_rate": 2.6688888888888894e-06,
-      "loss": 0.1912,
+      "loss": 0.4076,
       "step": 3800
     },
     {
-      "epoch": 2.443308847013734,
-      "grad_norm": 2.964355707168579,
+      "epoch": 2.5073746312684366,
+      "grad_norm": 5.018243312835693,
       "learning_rate": 2.6133333333333334e-06,
-      "loss": 0.1943,
+      "loss": 0.4095,
       "step": 3825
     },
     {
-      "epoch": 2.4592781858831043,
-      "grad_norm": 2.4776132106781006,
+      "epoch": 2.523762700753851,
+      "grad_norm": 5.116067886352539,
       "learning_rate": 2.557777777777778e-06,
-      "loss": 0.1916,
+      "loss": 0.419,
       "step": 3850
     },
     {
-      "epoch": 2.4752475247524752,
-      "grad_norm": 2.556891441345215,
+      "epoch": 2.540150770239266,
+      "grad_norm": 5.393177032470703,
       "learning_rate": 2.5022222222222224e-06,
-      "loss": 0.2041,
+      "loss": 0.4002,
       "step": 3875
     },
     {
-      "epoch": 2.491216863621846,
-      "grad_norm": 2.0888752937316895,
+      "epoch": 2.5565388397246807,
+      "grad_norm": 4.718799114227295,
       "learning_rate": 2.446666666666667e-06,
-      "loss": 0.2026,
+      "loss": 0.3743,
       "step": 3900
     },
     {
-      "epoch": 2.5071862024912166,
-      "grad_norm": 2.438257932662964,
+      "epoch": 2.572926909210095,
+      "grad_norm": 4.6258063316345215,
       "learning_rate": 2.3911111111111113e-06,
-      "loss": 0.2017,
+      "loss": 0.4004,
       "step": 3925
     },
     {
-      "epoch": 2.5231555413605875,
-      "grad_norm": 2.6804213523864746,
+      "epoch": 2.5893149786955094,
+      "grad_norm": 3.5877041816711426,
       "learning_rate": 2.3355555555555557e-06,
-      "loss": 0.1957,
+      "loss": 0.3895,
       "step": 3950
     },
     {
-      "epoch": 2.5391248802299584,
-      "grad_norm": 2.409219264984131,
+      "epoch": 2.6057030481809242,
+      "grad_norm": 4.622426509857178,
       "learning_rate": 2.28e-06,
-      "loss": 0.201,
+      "loss": 0.3799,
       "step": 3975
     },
     {
-      "epoch": 2.5550942190993293,
-      "grad_norm": 2.4475369453430176,
+      "epoch": 2.622091117666339,
+      "grad_norm": 4.797657489776611,
       "learning_rate": 2.2244444444444447e-06,
-      "loss": 0.1901,
+      "loss": 0.3903,
       "step": 4000
     },
     {
-      "epoch": 2.5550942190993293,
-      "eval_loss": 0.392954021692276,
-      "eval_runtime": 1661.0812,
-      "eval_samples_per_second": 2.348,
-      "eval_steps_per_second": 0.147,
-      "eval_wer": 0.2848932676518883,
+      "epoch": 2.622091117666339,
+      "eval_loss": 0.3909289240837097,
+      "eval_runtime": 1623.4733,
+      "eval_samples_per_second": 2.403,
+      "eval_steps_per_second": 0.15,
+      "eval_wer": 0.28537621945329855,
       "step": 4000
     },
     {
-      "epoch": 2.5710635579687002,
-      "grad_norm": 2.4416608810424805,
+      "epoch": 2.6384791871517534,
+      "grad_norm": 5.209569454193115,
       "learning_rate": 2.168888888888889e-06,
-      "loss": 0.1886,
+      "loss": 0.3973,
       "step": 4025
     },
     {
-      "epoch": 2.587032896838071,
-      "grad_norm": 2.4266738891601562,
+      "epoch": 2.6548672566371683,
+      "grad_norm": 5.018106937408447,
       "learning_rate": 2.1133333333333336e-06,
-      "loss": 0.1964,
+      "loss": 0.3867,
       "step": 4050
     },
     {
-      "epoch": 2.6030022357074416,
-      "grad_norm": 2.2254934310913086,
+      "epoch": 2.6712553261225827,
+      "grad_norm": 5.315625190734863,
       "learning_rate": 2.057777777777778e-06,
-      "loss": 0.193,
+      "loss": 0.397,
       "step": 4075
     },
     {
-      "epoch": 2.6189715745768125,
-      "grad_norm": 2.269014596939087,
+      "epoch": 2.6876433956079975,
+      "grad_norm": 4.67927885055542,
       "learning_rate": 2.0022222222222225e-06,
-      "loss": 0.1911,
+      "loss": 0.3796,
       "step": 4100
     },
     {
-      "epoch": 2.6349409134461834,
-      "grad_norm": 2.8170831203460693,
+      "epoch": 2.704031465093412,
+      "grad_norm": 4.914303779602051,
       "learning_rate": 1.9466666666666665e-06,
-      "loss": 0.1891,
+      "loss": 0.3856,
       "step": 4125
     },
     {
-      "epoch": 2.650910252315554,
-      "grad_norm": 2.5246517658233643,
+      "epoch": 2.7204195345788267,
+      "grad_norm": 5.619119167327881,
       "learning_rate": 1.8911111111111114e-06,
-      "loss": 0.2021,
+      "loss": 0.3957,
       "step": 4150
     },
     {
-      "epoch": 2.666879591184925,
-      "grad_norm": 2.7256410121917725,
+      "epoch": 2.736807604064241,
+      "grad_norm": 6.019058704376221,
       "learning_rate": 1.8355555555555557e-06,
-      "loss": 0.1924,
+      "loss": 0.3808,
       "step": 4175
     },
     {
-      "epoch": 2.6828489300542957,
-      "grad_norm": 2.446582555770874,
+      "epoch": 2.753195673549656,
+      "grad_norm": 4.6557745933532715,
       "learning_rate": 1.7800000000000001e-06,
-      "loss": 0.1934,
+      "loss": 0.3856,
       "step": 4200
     },
     {
-      "epoch": 2.6988182689236666,
-      "grad_norm": 2.602579116821289,
+      "epoch": 2.7695837430350707,
+      "grad_norm": 4.5428385734558105,
       "learning_rate": 1.7244444444444448e-06,
-      "loss": 0.188,
+      "loss": 0.3928,
       "step": 4225
     },
     {
-      "epoch": 2.7147876077930375,
-      "grad_norm": 2.2500104904174805,
+      "epoch": 2.785971812520485,
+      "grad_norm": 5.191465377807617,
       "learning_rate": 1.668888888888889e-06,
-      "loss": 0.1951,
+      "loss": 0.3821,
       "step": 4250
     },
     {
-      "epoch": 2.7307569466624084,
-      "grad_norm": 2.7052624225616455,
+      "epoch": 2.8023598820058995,
+      "grad_norm": 4.012134552001953,
       "learning_rate": 1.6133333333333335e-06,
-      "loss": 0.2039,
+      "loss": 0.3945,
       "step": 4275
     },
     {
-      "epoch": 2.746726285531779,
-      "grad_norm": 2.827786922454834,
+      "epoch": 2.8187479514913143,
+      "grad_norm": 6.12072229385376,
       "learning_rate": 1.5577777777777777e-06,
-      "loss": 0.1859,
+      "loss": 0.4265,
       "step": 4300
     },
     {
-      "epoch": 2.76269562440115,
-      "grad_norm": 2.8331093788146973,
+      "epoch": 2.835136020976729,
+      "grad_norm": 5.369980812072754,
       "learning_rate": 1.5022222222222224e-06,
-      "loss": 0.2028,
+      "loss": 0.3891,
       "step": 4325
     },
     {
-      "epoch": 2.7786649632705207,
-      "grad_norm": 2.4881205558776855,
+      "epoch": 2.8515240904621435,
+      "grad_norm": 5.052566051483154,
       "learning_rate": 1.4466666666666669e-06,
-      "loss": 0.2013,
+      "loss": 0.3855,
       "step": 4350
     },
     {
-      "epoch": 2.794634302139891,
-      "grad_norm": 2.725170612335205,
+      "epoch": 2.8679121599475583,
+      "grad_norm": 4.885142803192139,
       "learning_rate": 1.3911111111111111e-06,
-      "loss": 0.202,
+      "loss": 0.3941,
       "step": 4375
     },
     {
-      "epoch": 2.810603641009262,
-      "grad_norm": 2.9598212242126465,
+      "epoch": 2.8843002294329727,
+      "grad_norm": 5.071890354156494,
       "learning_rate": 1.3355555555555558e-06,
-      "loss": 0.1957,
+      "loss": 0.3907,
       "step": 4400
     },
     {
-      "epoch": 2.826572979878633,
-      "grad_norm": 2.1554248332977295,
+      "epoch": 2.9006882989183875,
+      "grad_norm": 7.670461177825928,
       "learning_rate": 1.28e-06,
-      "loss": 0.185,
+      "loss": 0.3602,
       "step": 4425
     },
     {
-      "epoch": 2.842542318748004,
-      "grad_norm": 2.447131872177124,
+      "epoch": 2.917076368403802,
+      "grad_norm": 5.3365373611450195,
       "learning_rate": 1.2244444444444445e-06,
-      "loss": 0.197,
+      "loss": 0.3766,
       "step": 4450
     },
     {
-      "epoch": 2.858511657617375,
-      "grad_norm": 2.1197404861450195,
+      "epoch": 2.9334644378892167,
+      "grad_norm": 4.464676856994629,
       "learning_rate": 1.168888888888889e-06,
-      "loss": 0.189,
+      "loss": 0.3768,
       "step": 4475
     },
     {
-      "epoch": 2.8744809964867457,
-      "grad_norm": 2.2641892433166504,
+      "epoch": 2.949852507374631,
+      "grad_norm": 5.319380760192871,
       "learning_rate": 1.1133333333333334e-06,
-      "loss": 0.1825,
+      "loss": 0.3646,
       "step": 4500
     },
     {
-      "epoch": 2.890450335356116,
-      "grad_norm": 2.380861759185791,
+      "epoch": 2.966240576860046,
+      "grad_norm": 5.396827697753906,
       "learning_rate": 1.0577777777777779e-06,
-      "loss": 0.186,
+      "loss": 0.3813,
       "step": 4525
     },
     {
-      "epoch": 2.906419674225487,
-      "grad_norm": 2.575404167175293,
+      "epoch": 2.9826286463454608,
+      "grad_norm": 5.3341450691223145,
       "learning_rate": 1.0022222222222223e-06,
-      "loss": 0.194,
+      "loss": 0.4021,
       "step": 4550
     },
     {
-      "epoch": 2.922389013094858,
-      "grad_norm": 1.9462391138076782,
+      "epoch": 2.999016715830875,
+      "grad_norm": 4.901559352874756,
       "learning_rate": 9.466666666666667e-07,
-      "loss": 0.2058,
+      "loss": 0.3835,
       "step": 4575
     },
     {
-      "epoch": 2.9383583519642285,
-      "grad_norm": 2.332577705383301,
+      "epoch": 3.0154047853162895,
+      "grad_norm": 4.087602615356445,
       "learning_rate": 8.911111111111112e-07,
-      "loss": 0.1864,
+      "loss": 0.2653,
       "step": 4600
     },
     {
-      "epoch": 2.9543276908335994,
-      "grad_norm": 2.345797300338745,
+      "epoch": 3.0317928548017044,
+      "grad_norm": 4.191402435302734,
       "learning_rate": 8.355555555555556e-07,
-      "loss": 0.1955,
+      "loss": 0.2492,
       "step": 4625
     },
     {
-      "epoch": 2.9702970297029703,
-      "grad_norm": 2.41668438911438,
+      "epoch": 3.048180924287119,
+      "grad_norm": 4.047247409820557,
       "learning_rate": 7.8e-07,
-      "loss": 0.1804,
+      "loss": 0.2529,
       "step": 4650
     },
     {
-      "epoch": 2.986266368572341,
-      "grad_norm": 2.3288943767547607,
+      "epoch": 3.0645689937725336,
+      "grad_norm": 6.462422847747803,
       "learning_rate": 7.244444444444446e-07,
-      "loss": 0.1911,
+      "loss": 0.257,
       "step": 4675
     },
     {
-      "epoch": 3.002235707441712,
-      "grad_norm": 1.3726705312728882,
+      "epoch": 3.0809570632579484,
+      "grad_norm": 4.2245774269104,
       "learning_rate": 6.68888888888889e-07,
-      "loss": 0.1877,
+      "loss": 0.2618,
       "step": 4700
     },
     {
-      "epoch": 3.0182050463110826,
-      "grad_norm": 2.3931479454040527,
+      "epoch": 3.0973451327433628,
+      "grad_norm": 3.604339838027954,
       "learning_rate": 6.133333333333333e-07,
-      "loss": 0.1261,
+      "loss": 0.2774,
       "step": 4725
     },
     {
-      "epoch": 3.0341743851804535,
-      "grad_norm": 1.6263169050216675,
+      "epoch": 3.1137332022287776,
+      "grad_norm": 3.8819730281829834,
       "learning_rate": 5.577777777777779e-07,
-      "loss": 0.1247,
+      "loss": 0.2618,
       "step": 4750
     },
     {
-      "epoch": 3.0501437240498244,
-      "grad_norm": 1.9789073467254639,
+      "epoch": 3.130121271714192,
+      "grad_norm": 4.699184417724609,
       "learning_rate": 5.022222222222222e-07,
-      "loss": 0.1259,
+      "loss": 0.2591,
       "step": 4775
     },
     {
-      "epoch": 3.0661130629191953,
-      "grad_norm": 2.2936391830444336,
+      "epoch": 3.146509341199607,
+      "grad_norm": 4.058821201324463,
       "learning_rate": 4.466666666666667e-07,
-      "loss": 0.1318,
+      "loss": 0.2735,
       "step": 4800
     },
     {
-      "epoch": 3.0820824017885657,
-      "grad_norm": 2.009859800338745,
+      "epoch": 3.162897410685021,
+      "grad_norm": 3.7905142307281494,
       "learning_rate": 3.9111111111111115e-07,
-      "loss": 0.1346,
+      "loss": 0.2648,
       "step": 4825
     },
     {
-      "epoch": 3.0980517406579366,
-      "grad_norm": 2.502535104751587,
+      "epoch": 3.179285480170436,
+      "grad_norm": 5.008088111877441,
       "learning_rate": 3.3555555555555556e-07,
-      "loss": 0.1312,
+      "loss": 0.2646,
       "step": 4850
     },
     {
-      "epoch": 3.1140210795273076,
-      "grad_norm": 2.6293087005615234,
+      "epoch": 3.1956735496558504,
+      "grad_norm": 4.818483829498291,
       "learning_rate": 2.8e-07,
-      "loss": 0.1277,
+      "loss": 0.264,
       "step": 4875
     },
     {
-      "epoch": 3.1299904183966785,
-      "grad_norm": 2.0969161987304688,
+      "epoch": 3.212061619141265,
+      "grad_norm": 3.8618814945220947,
       "learning_rate": 2.2444444444444445e-07,
-      "loss": 0.1301,
+      "loss": 0.255,
       "step": 4900
     },
     {
-      "epoch": 3.1459597572660494,
-      "grad_norm": 1.9938933849334717,
+      "epoch": 3.2284496886266796,
+      "grad_norm": 4.410851955413818,
       "learning_rate": 1.6888888888888888e-07,
-      "loss": 0.129,
+      "loss": 0.2538,
       "step": 4925
     },
     {
-      "epoch": 3.16192909613542,
-      "grad_norm": 2.1169307231903076,
+      "epoch": 3.2448377581120944,
+      "grad_norm": 4.055325508117676,
       "learning_rate": 1.1333333333333336e-07,
-      "loss": 0.134,
+      "loss": 0.2506,
       "step": 4950
     },
     {
-      "epoch": 3.1778984350047907,
-      "grad_norm": 2.5867435932159424,
+      "epoch": 3.2612258275975092,
+      "grad_norm": 3.297287940979004,
       "learning_rate": 5.777777777777778e-08,
-      "loss": 0.1379,
+      "loss": 0.2471,
       "step": 4975
     },
     {
-      "epoch": 3.1938677738741617,
-      "grad_norm": 2.1530654430389404,
+      "epoch": 3.2776138970829236,
+      "grad_norm": 3.850027084350586,
       "learning_rate": 2.2222222222222225e-09,
-      "loss": 0.1306,
+      "loss": 0.2479,
       "step": 5000
     },
     {
-      "epoch": 3.1938677738741617,
-      "eval_loss": 0.40369176864624023,
-      "eval_runtime": 1644.4554,
-      "eval_samples_per_second": 2.372,
-      "eval_steps_per_second": 0.148,
-      "eval_wer": 0.27909784603496574,
+      "epoch": 3.2776138970829236,
+      "eval_loss": 0.40801551938056946,
+      "eval_runtime": 1621.9704,
+      "eval_samples_per_second": 2.405,
+      "eval_steps_per_second": 0.15,
+      "eval_wer": 0.2782285327924273,
       "step": 5000
     },
     {
-      "epoch": 3.1938677738741617,
+      "epoch": 3.2776138970829236,
       "step": 5000,
       "total_flos": 5.435895365546803e+20,
-      "train_loss": 0.3537434986591339,
-      "train_runtime": 57744.1557,
-      "train_samples_per_second": 2.771,
-      "train_steps_per_second": 0.087
+      "train_loss": 0.6995921184539795,
+      "train_runtime": 59796.6444,
+      "train_samples_per_second": 2.676,
+      "train_steps_per_second": 0.084
     }
   ],
   "logging_steps": 25,