End of training

README.md

@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-large-v3-ft-cv-cy-en
 
-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/commonvoice_18_0_cy_en train main dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.3036
 - Wer: 0.1826

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 10.030090270812437,
-    "eval_loss": 0.4447694718837738,
-    "eval_runtime": 4494.8849,
+    "epoch": 3.537318712415989,
+    "eval_loss": 0.3035752773284912,
+    "eval_runtime": 4625.832,
     "eval_samples": 10763,
-    "eval_samples_per_second": 2.394,
-    "eval_steps_per_second": 0.15,
-    "eval_wer": 0.1960611749980443,
-    "total_flos": 5.43090104303616e+20,
-    "train_loss": 0.0887994671344757,
-    "train_runtime": 50408.8653,
-    "train_samples": 15937,
-    "train_samples_per_second": 3.174,
-    "train_steps_per_second": 0.099
+    "eval_samples_per_second": 2.327,
+    "eval_steps_per_second": 0.145,
+    "eval_wer": 0.18264491903309082,
+    "total_flos": 5.435079965953229e+20,
+    "train_loss": 0.17054095619916915,
+    "train_runtime": 54364.2112,
+    "train_samples": 45223,
+    "train_samples_per_second": 2.943,
+    "train_steps_per_second": 0.092
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 10.030090270812437,
-    "eval_loss": 0.4447694718837738,
-    "eval_runtime": 4494.8849,
+    "epoch": 3.537318712415989,
+    "eval_loss": 0.3035752773284912,
+    "eval_runtime": 4625.832,
     "eval_samples": 10763,
-    "eval_samples_per_second": 2.394,
-    "eval_steps_per_second": 0.15,
-    "eval_wer": 0.1960611749980443
+    "eval_samples_per_second": 2.327,
+    "eval_steps_per_second": 0.145,
+    "eval_wer": 0.18264491903309082
 }

runs/Oct21_11-04-11_712574289d1f/events.out.tfevents.1729565286.712574289d1f.141.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f12299f5f1b92917190d7c55067875e5b05c6f423c1856318c69c758999da2b3
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 10.030090270812437,
-    "total_flos": 5.43090104303616e+20,
-    "train_loss": 0.0887994671344757,
-    "train_runtime": 50408.8653,
-    "train_samples": 15937,
-    "train_samples_per_second": 3.174,
-    "train_steps_per_second": 0.099
+    "epoch": 3.537318712415989,
+    "total_flos": 5.435079965953229e+20,
+    "train_loss": 0.17054095619916915,
+    "train_runtime": 54364.2112,
+    "train_samples": 45223,
+    "train_samples_per_second": 2.943,
+    "train_steps_per_second": 0.092
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 10.030090270812437,
+  "epoch": 3.537318712415989,
   "eval_steps": 1000,
   "global_step": 5000,
   "is_hyper_param_search": false,
@@ -9,1464 +9,1464 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.05015045135406219,
-      "grad_norm": 7.437633514404297,
+      "epoch": 0.017686593562079942,
+      "grad_norm": 4.990772724151611,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.9756,
+      "loss": 0.9549,
       "step": 25
     },
     {
-      "epoch": 0.10030090270812438,
-      "grad_norm": 3.9726901054382324,
+      "epoch": 0.035373187124159884,
+      "grad_norm": 4.042325019836426,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.8322,
+      "loss": 0.8294,
       "step": 50
     },
     {
-      "epoch": 0.15045135406218657,
-      "grad_norm": 3.629199743270874,
+      "epoch": 0.05305978068623983,
+      "grad_norm": 3.429729461669922,
       "learning_rate": 1.5e-06,
-      "loss": 0.5489,
+      "loss": 0.5478,
       "step": 75
     },
     {
-      "epoch": 0.20060180541624875,
-      "grad_norm": 3.2304935455322266,
+      "epoch": 0.07074637424831977,
+      "grad_norm": 3.084613800048828,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.4407,
+      "loss": 0.482,
       "step": 100
     },
     {
-      "epoch": 0.25075225677031093,
-      "grad_norm": 3.2816317081451416,
+      "epoch": 0.08843296781039972,
+      "grad_norm": 3.2256970405578613,
       "learning_rate": 2.5e-06,
-      "loss": 0.4424,
+      "loss": 0.4206,
       "step": 125
     },
     {
-      "epoch": 0.30090270812437314,
-      "grad_norm": 3.1936333179473877,
+      "epoch": 0.10611956137247966,
+      "grad_norm": 3.0460891723632812,
       "learning_rate": 3e-06,
-      "loss": 0.4074,
+      "loss": 0.4171,
       "step": 150
     },
     {
-      "epoch": 0.3510531594784353,
-      "grad_norm": 3.1005301475524902,
+      "epoch": 0.1238061549345596,
+      "grad_norm": 2.976763963699341,
       "learning_rate": 3.5e-06,
-      "loss": 0.4104,
+      "loss": 0.3896,
       "step": 175
     },
     {
-      "epoch": 0.4012036108324975,
-      "grad_norm": 3.329033851623535,
+      "epoch": 0.14149274849663954,
+      "grad_norm": 2.8915836811065674,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.4049,
+      "loss": 0.4149,
       "step": 200
     },
     {
-      "epoch": 0.45135406218655966,
-      "grad_norm": 2.9963624477386475,
+      "epoch": 0.1591793420587195,
+      "grad_norm": 3.01548433303833,
       "learning_rate": 4.5e-06,
-      "loss": 0.3773,
+      "loss": 0.3811,
       "step": 225
     },
     {
-      "epoch": 0.5015045135406219,
-      "grad_norm": 2.877650260925293,
+      "epoch": 0.17686593562079944,
+      "grad_norm": 3.8455846309661865,
       "learning_rate": 5e-06,
-      "loss": 0.3659,
+      "loss": 0.3872,
       "step": 250
     },
     {
-      "epoch": 0.551654964894684,
-      "grad_norm": 3.1805803775787354,
+      "epoch": 0.19455252918287938,
+      "grad_norm": 2.805778741836548,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.3686,
+      "loss": 0.3714,
       "step": 275
     },
     {
-      "epoch": 0.6018054162487463,
-      "grad_norm": 2.518958806991577,
+      "epoch": 0.21223912274495932,
+      "grad_norm": 3.138227939605713,
       "learning_rate": 6e-06,
-      "loss": 0.362,
+      "loss": 0.3768,
       "step": 300
     },
     {
-      "epoch": 0.6519558676028084,
-      "grad_norm": 2.588172674179077,
+      "epoch": 0.22992571630703926,
+      "grad_norm": 3.5028634071350098,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.3603,
+      "loss": 0.3835,
       "step": 325
     },
     {
-      "epoch": 0.7021063189568706,
-      "grad_norm": 3.022862672805786,
+      "epoch": 0.2476123098691192,
+      "grad_norm": 3.8886911869049072,
       "learning_rate": 7e-06,
-      "loss": 0.3835,
+      "loss": 0.359,
       "step": 350
     },
     {
-      "epoch": 0.7522567703109327,
-      "grad_norm": 2.3474695682525635,
+      "epoch": 0.26529890343119916,
+      "grad_norm": 2.8521852493286133,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.3567,
+      "loss": 0.3507,
       "step": 375
     },
     {
-      "epoch": 0.802407221664995,
-      "grad_norm": 2.847740411758423,
+      "epoch": 0.2829854969932791,
+      "grad_norm": 2.901258945465088,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.3566,
+      "loss": 0.3586,
       "step": 400
     },
     {
-      "epoch": 0.8525576730190572,
-      "grad_norm": 2.8755927085876465,
+      "epoch": 0.30067209055535904,
+      "grad_norm": 2.550757884979248,
       "learning_rate": 8.5e-06,
-      "loss": 0.3552,
+      "loss": 0.3658,
       "step": 425
     },
     {
-      "epoch": 0.9027081243731193,
-      "grad_norm": 2.488140106201172,
+      "epoch": 0.318358684117439,
+      "grad_norm": 3.0283052921295166,
       "learning_rate": 9e-06,
-      "loss": 0.3574,
+      "loss": 0.3527,
       "step": 450
     },
     {
-      "epoch": 0.9528585757271816,
-      "grad_norm": 3.7667624950408936,
+      "epoch": 0.3360452776795189,
+      "grad_norm": 3.272336721420288,
       "learning_rate": 9.5e-06,
-      "loss": 0.34,
+      "loss": 0.3674,
       "step": 475
     },
     {
-      "epoch": 1.0030090270812437,
-      "grad_norm": 2.2906906604766846,
+      "epoch": 0.3537318712415989,
+      "grad_norm": 2.8109354972839355,
       "learning_rate": 1e-05,
-      "loss": 0.332,
+      "loss": 0.3592,
       "step": 500
     },
     {
-      "epoch": 1.053159478435306,
-      "grad_norm": 2.970755100250244,
+      "epoch": 0.3714184648036788,
+      "grad_norm": 2.21569561958313,
       "learning_rate": 9.944444444444445e-06,
-      "loss": 0.245,
+      "loss": 0.3219,
       "step": 525
     },
     {
-      "epoch": 1.103309929789368,
-      "grad_norm": 2.735118865966797,
+      "epoch": 0.38910505836575876,
+      "grad_norm": 2.5520081520080566,
       "learning_rate": 9.88888888888889e-06,
-      "loss": 0.2632,
+      "loss": 0.3681,
       "step": 550
     },
     {
-      "epoch": 1.1534603811434303,
-      "grad_norm": 2.1645102500915527,
+      "epoch": 0.40679165192783867,
+      "grad_norm": 2.903778076171875,
       "learning_rate": 9.833333333333333e-06,
-      "loss": 0.2485,
+      "loss": 0.3424,
       "step": 575
     },
     {
-      "epoch": 1.2036108324974926,
-      "grad_norm": 2.3044114112854004,
+      "epoch": 0.42447824548991864,
+      "grad_norm": 2.4832980632781982,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 0.2545,
+      "loss": 0.3269,
       "step": 600
     },
     {
-      "epoch": 1.2537612838515546,
-      "grad_norm": 2.445683240890503,
+      "epoch": 0.4421648390519986,
+      "grad_norm": 2.759129285812378,
       "learning_rate": 9.722222222222223e-06,
-      "loss": 0.2398,
+      "loss": 0.327,
       "step": 625
     },
     {
-      "epoch": 1.3039117352056169,
-      "grad_norm": 1.9902832508087158,
+      "epoch": 0.4598514326140785,
+      "grad_norm": 2.3846707344055176,
       "learning_rate": 9.666666666666667e-06,
-      "loss": 0.2374,
+      "loss": 0.3367,
       "step": 650
     },
     {
-      "epoch": 1.3540621865596791,
-      "grad_norm": 2.571229934692383,
+      "epoch": 0.4775380261761585,
+      "grad_norm": 2.462186574935913,
       "learning_rate": 9.611111111111112e-06,
-      "loss": 0.2532,
+      "loss": 0.3175,
       "step": 675
     },
     {
-      "epoch": 1.4042126379137412,
-      "grad_norm": 2.24031662940979,
+      "epoch": 0.4952246197382384,
+      "grad_norm": 2.5651514530181885,
       "learning_rate": 9.555555555555556e-06,
-      "loss": 0.2466,
+      "loss": 0.3156,
       "step": 700
     },
     {
-      "epoch": 1.4543630892678034,
-      "grad_norm": 2.2203283309936523,
+      "epoch": 0.5129112133003184,
+      "grad_norm": 2.6540584564208984,
       "learning_rate": 9.5e-06,
-      "loss": 0.2414,
+      "loss": 0.3098,
       "step": 725
     },
     {
-      "epoch": 1.5045135406218657,
-      "grad_norm": 2.821672201156616,
+      "epoch": 0.5305978068623983,
+      "grad_norm": 2.713265895843506,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 0.2612,
+      "loss": 0.3226,
       "step": 750
     },
     {
-      "epoch": 1.5546639919759278,
-      "grad_norm": 2.279815196990967,
+      "epoch": 0.5482844004244782,
+      "grad_norm": 2.128851890563965,
       "learning_rate": 9.38888888888889e-06,
-      "loss": 0.2361,
+      "loss": 0.3244,
       "step": 775
     },
     {
-      "epoch": 1.60481444332999,
-      "grad_norm": 2.2528128623962402,
+      "epoch": 0.5659709939865581,
+      "grad_norm": 2.5764191150665283,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 0.2337,
+      "loss": 0.3156,
       "step": 800
     },
     {
-      "epoch": 1.6549648946840523,
-      "grad_norm": 2.5302581787109375,
+      "epoch": 0.5836575875486382,
+      "grad_norm": 2.1402037143707275,
       "learning_rate": 9.277777777777778e-06,
-      "loss": 0.2407,
+      "loss": 0.3029,
       "step": 825
     },
     {
-      "epoch": 1.7051153460381143,
-      "grad_norm": 2.0469911098480225,
+      "epoch": 0.6013441811107181,
+      "grad_norm": 2.4773480892181396,
       "learning_rate": 9.222222222222224e-06,
-      "loss": 0.2603,
+      "loss": 0.293,
       "step": 850
     },
     {
-      "epoch": 1.7552657973921764,
-      "grad_norm": 2.5774879455566406,
+      "epoch": 0.619030774672798,
+      "grad_norm": 2.6644790172576904,
       "learning_rate": 9.166666666666666e-06,
-      "loss": 0.2441,
+      "loss": 0.3111,
       "step": 875
     },
     {
-      "epoch": 1.8054162487462388,
-      "grad_norm": 2.362243175506592,
+      "epoch": 0.636717368234878,
+      "grad_norm": 2.3241961002349854,
       "learning_rate": 9.111111111111112e-06,
-      "loss": 0.2315,
+      "loss": 0.2726,
       "step": 900
     },
     {
-      "epoch": 1.855566700100301,
-      "grad_norm": 2.5620100498199463,
+      "epoch": 0.6544039617969579,
+      "grad_norm": 2.6867661476135254,
       "learning_rate": 9.055555555555556e-06,
-      "loss": 0.2388,
+      "loss": 0.3083,
       "step": 925
     },
     {
-      "epoch": 1.905717151454363,
-      "grad_norm": 2.1473169326782227,
+      "epoch": 0.6720905553590378,
+      "grad_norm": 2.138195037841797,
       "learning_rate": 9e-06,
-      "loss": 0.23,
+      "loss": 0.3019,
       "step": 950
     },
     {
-      "epoch": 1.9558676028084254,
-      "grad_norm": 2.693861722946167,
+      "epoch": 0.6897771489211177,
+      "grad_norm": 2.6360018253326416,
       "learning_rate": 8.944444444444446e-06,
-      "loss": 0.2491,
+      "loss": 0.2917,
       "step": 975
     },
     {
-      "epoch": 2.0060180541624875,
-      "grad_norm": 1.6720950603485107,
+      "epoch": 0.7074637424831978,
+      "grad_norm": 2.5191752910614014,
       "learning_rate": 8.888888888888888e-06,
-      "loss": 0.2144,
+      "loss": 0.294,
       "step": 1000
     },
     {
-      "epoch": 2.0060180541624875,
-      "eval_loss": 0.3075518310070038,
-      "eval_runtime": 4546.1548,
-      "eval_samples_per_second": 2.367,
-      "eval_steps_per_second": 0.148,
-      "eval_wer": 0.2114135961824298,
+      "epoch": 0.7074637424831978,
+      "eval_loss": 0.30541032552719116,
+      "eval_runtime": 4620.7401,
+      "eval_samples_per_second": 2.329,
+      "eval_steps_per_second": 0.146,
+      "eval_wer": 0.21139403895799108,
       "step": 1000
     },
     {
-      "epoch": 2.0561685055165495,
-      "grad_norm": 2.0117599964141846,
+      "epoch": 0.7251503360452777,
+      "grad_norm": 2.696348190307617,
       "learning_rate": 8.833333333333334e-06,
-      "loss": 0.1061,
+      "loss": 0.2912,
       "step": 1025
     },
     {
-      "epoch": 2.106318956870612,
-      "grad_norm": 1.9226505756378174,
+      "epoch": 0.7428369296073576,
+      "grad_norm": 2.3700759410858154,
       "learning_rate": 8.777777777777778e-06,
-      "loss": 0.1171,
+      "loss": 0.2756,
       "step": 1050
     },
     {
-      "epoch": 2.156469408224674,
-      "grad_norm": 2.096928358078003,
+      "epoch": 0.7605235231694376,
+      "grad_norm": 2.314389705657959,
       "learning_rate": 8.722222222222224e-06,
-      "loss": 0.1135,
+      "loss": 0.2881,
       "step": 1075
     },
     {
-      "epoch": 2.206619859578736,
-      "grad_norm": 1.7984790802001953,
+      "epoch": 0.7782101167315175,
+      "grad_norm": 2.3540844917297363,
       "learning_rate": 8.666666666666668e-06,
-      "loss": 0.1164,
+      "loss": 0.291,
       "step": 1100
     },
     {
-      "epoch": 2.2567703109327986,
-      "grad_norm": 1.6253271102905273,
+      "epoch": 0.7958967102935974,
+      "grad_norm": 2.3578970432281494,
       "learning_rate": 8.611111111111112e-06,
-      "loss": 0.1136,
+      "loss": 0.2846,
       "step": 1125
     },
     {
-      "epoch": 2.3069207622868606,
-      "grad_norm": 1.7676957845687866,
+      "epoch": 0.8135833038556773,
+      "grad_norm": 1.976126790046692,
       "learning_rate": 8.555555555555556e-06,
-      "loss": 0.1207,
+      "loss": 0.27,
       "step": 1150
     },
     {
-      "epoch": 2.3570712136409226,
-      "grad_norm": 2.040688991546631,
+      "epoch": 0.8312698974177574,
+      "grad_norm": 2.194822072982788,
       "learning_rate": 8.5e-06,
-      "loss": 0.1148,
+      "loss": 0.2763,
       "step": 1175
     },
     {
-      "epoch": 2.407221664994985,
-      "grad_norm": 2.272294282913208,
+      "epoch": 0.8489564909798373,
+      "grad_norm": 2.1482298374176025,
       "learning_rate": 8.444444444444446e-06,
-      "loss": 0.1222,
+      "loss": 0.2731,
       "step": 1200
     },
     {
-      "epoch": 2.457372116349047,
-      "grad_norm": 2.0784871578216553,
+      "epoch": 0.8666430845419172,
+      "grad_norm": 2.163198471069336,
       "learning_rate": 8.38888888888889e-06,
-      "loss": 0.1188,
+      "loss": 0.2602,
       "step": 1225
     },
     {
-      "epoch": 2.507522567703109,
-      "grad_norm": 1.869183897972107,
+      "epoch": 0.8843296781039972,
+      "grad_norm": 2.482328414916992,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.1187,
+      "loss": 0.2896,
       "step": 1250
     },
     {
-      "epoch": 2.5576730190571713,
-      "grad_norm": 2.0832390785217285,
+      "epoch": 0.9020162716660771,
+      "grad_norm": 2.2900679111480713,
       "learning_rate": 8.277777777777778e-06,
-      "loss": 0.1176,
+      "loss": 0.2472,
       "step": 1275
     },
     {
-      "epoch": 2.6078234704112337,
-      "grad_norm": 1.9388813972473145,
+      "epoch": 0.919702865228157,
+      "grad_norm": 2.4348435401916504,
       "learning_rate": 8.222222222222222e-06,
-      "loss": 0.1146,
+      "loss": 0.2754,
       "step": 1300
     },
     {
-      "epoch": 2.657973921765296,
-      "grad_norm": 1.659645676612854,
+      "epoch": 0.937389458790237,
+      "grad_norm": 2.4213294982910156,
       "learning_rate": 8.166666666666668e-06,
-      "loss": 0.116,
+      "loss": 0.2585,
       "step": 1325
     },
     {
-      "epoch": 2.7081243731193583,
-      "grad_norm": 1.7133307456970215,
+      "epoch": 0.955076052352317,
+      "grad_norm": 2.6255362033843994,
       "learning_rate": 8.111111111111112e-06,
-      "loss": 0.1201,
+      "loss": 0.25,
       "step": 1350
     },
     {
-      "epoch": 2.7582748244734203,
-      "grad_norm": 2.0739212036132812,
+      "epoch": 0.9727626459143969,
+      "grad_norm": 2.187410354614258,
       "learning_rate": 8.055555555555557e-06,
-      "loss": 0.1177,
+      "loss": 0.2809,
       "step": 1375
     },
     {
-      "epoch": 2.8084252758274824,
-      "grad_norm": 1.760362148284912,
+      "epoch": 0.9904492394764768,
+      "grad_norm": 2.669609546661377,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.1071,
+      "loss": 0.2707,
       "step": 1400
     },
     {
-      "epoch": 2.8585757271815444,
-      "grad_norm": 1.979827880859375,
+      "epoch": 1.0081358330385568,
+      "grad_norm": 1.841189980506897,
       "learning_rate": 7.944444444444445e-06,
-      "loss": 0.1141,
+      "loss": 0.2272,
       "step": 1425
     },
     {
-      "epoch": 2.908726178535607,
-      "grad_norm": 1.8925753831863403,
+      "epoch": 1.0258224266006368,
+      "grad_norm": 1.8432203531265259,
       "learning_rate": 7.88888888888889e-06,
-      "loss": 0.1272,
+      "loss": 0.2036,
       "step": 1450
     },
     {
-      "epoch": 2.958876629889669,
-      "grad_norm": 1.9880926609039307,
+      "epoch": 1.0435090201627166,
+      "grad_norm": 1.961900234222412,
       "learning_rate": 7.833333333333333e-06,
-      "loss": 0.1189,
+      "loss": 0.1758,
       "step": 1475
     },
     {
-      "epoch": 3.0090270812437314,
-      "grad_norm": 1.2164592742919922,
+      "epoch": 1.0611956137247966,
+      "grad_norm": 1.7587320804595947,
       "learning_rate": 7.77777777777778e-06,
-      "loss": 0.1091,
+      "loss": 0.1847,
       "step": 1500
     },
     {
-      "epoch": 3.0591775325977935,
-      "grad_norm": 1.2302759885787964,
+      "epoch": 1.0788822072868764,
+      "grad_norm": 2.330167531967163,
       "learning_rate": 7.722222222222223e-06,
-      "loss": 0.0472,
+      "loss": 0.1602,
       "step": 1525
     },
     {
-      "epoch": 3.1093279839518555,
-      "grad_norm": 1.2639251947402954,
+      "epoch": 1.0965688008489565,
+      "grad_norm": 2.3043272495269775,
       "learning_rate": 7.666666666666667e-06,
-      "loss": 0.048,
+      "loss": 0.1772,
       "step": 1550
     },
     {
-      "epoch": 3.1594784353059175,
-      "grad_norm": 1.2657603025436401,
+      "epoch": 1.1142553944110365,
+      "grad_norm": 2.093520164489746,
       "learning_rate": 7.611111111111111e-06,
-      "loss": 0.0488,
+      "loss": 0.1742,
       "step": 1575
     },
     {
-      "epoch": 3.20962888665998,
-      "grad_norm": 1.3162034749984741,
+      "epoch": 1.1319419879731163,
+      "grad_norm": 2.2821807861328125,
       "learning_rate": 7.555555555555556e-06,
-      "loss": 0.0486,
+      "loss": 0.1685,
       "step": 1600
     },
     {
-      "epoch": 3.259779338014042,
-      "grad_norm": 1.4205294847488403,
+      "epoch": 1.1496285815351963,
+      "grad_norm": 2.087881326675415,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.0484,
+      "loss": 0.1722,
       "step": 1625
     },
     {
-      "epoch": 3.3099297893681046,
-      "grad_norm": 1.3729196786880493,
+      "epoch": 1.1673151750972763,
+      "grad_norm": 1.7071295976638794,
       "learning_rate": 7.444444444444445e-06,
-      "loss": 0.0512,
+      "loss": 0.168,
       "step": 1650
     },
     {
-      "epoch": 3.3600802407221666,
-      "grad_norm": 1.333731770515442,
+      "epoch": 1.1850017686593561,
+      "grad_norm": 2.3514926433563232,
       "learning_rate": 7.38888888888889e-06,
-      "loss": 0.05,
+      "loss": 0.1786,
       "step": 1675
     },
     {
-      "epoch": 3.4102306920762286,
-      "grad_norm": 1.2979406118392944,
+      "epoch": 1.2026883622214362,
+      "grad_norm": 2.0674052238464355,
       "learning_rate": 7.333333333333333e-06,
-      "loss": 0.0535,
+      "loss": 0.1625,
       "step": 1700
     },
     {
-      "epoch": 3.4603811434302907,
-      "grad_norm": 1.1790673732757568,
+      "epoch": 1.2203749557835162,
+      "grad_norm": 1.9389206171035767,
       "learning_rate": 7.277777777777778e-06,
-      "loss": 0.0598,
+      "loss": 0.1686,
       "step": 1725
     },
     {
-      "epoch": 3.510531594784353,
-      "grad_norm": 1.7529319524765015,
+      "epoch": 1.238061549345596,
+      "grad_norm": 2.185607433319092,
       "learning_rate": 7.222222222222223e-06,
-      "loss": 0.0539,
+      "loss": 0.168,
       "step": 1750
     },
     {
-      "epoch": 3.560682046138415,
-      "grad_norm": 1.5352625846862793,
+      "epoch": 1.255748142907676,
+      "grad_norm": 1.953815221786499,
       "learning_rate": 7.166666666666667e-06,
-      "loss": 0.0494,
+      "loss": 0.1634,
       "step": 1775
     },
     {
-      "epoch": 3.6108324974924777,
-      "grad_norm": 1.707581639289856,
+      "epoch": 1.2734347364697558,
+      "grad_norm": 2.0842204093933105,
       "learning_rate": 7.111111111111112e-06,
-      "loss": 0.051,
+      "loss": 0.1749,
       "step": 1800
     },
     {
-      "epoch": 3.6609829488465397,
-      "grad_norm": 1.3295681476593018,
+      "epoch": 1.2911213300318358,
+      "grad_norm": 2.197368621826172,
       "learning_rate": 7.055555555555557e-06,
-      "loss": 0.0545,
+      "loss": 0.173,
       "step": 1825
     },
     {
-      "epoch": 3.711133400200602,
-      "grad_norm": 1.3010529279708862,
+      "epoch": 1.3088079235939158,
+      "grad_norm": 1.8195652961730957,
       "learning_rate": 7e-06,
-      "loss": 0.0534,
+      "loss": 0.1575,
       "step": 1850
     },
     {
-      "epoch": 3.761283851554664,
-      "grad_norm": 1.0611276626586914,
+      "epoch": 1.3264945171559956,
+      "grad_norm": 1.9379323720932007,
       "learning_rate": 6.944444444444445e-06,
-      "loss": 0.0509,
+      "loss": 0.1591,
       "step": 1875
     },
     {
-      "epoch": 3.8114343029087263,
-      "grad_norm": 1.1071949005126953,
+      "epoch": 1.3441811107180757,
+      "grad_norm": 1.8081250190734863,
       "learning_rate": 6.88888888888889e-06,
-      "loss": 0.0496,
+      "loss": 0.1534,
       "step": 1900
     },
     {
-      "epoch": 3.8615847542627884,
-      "grad_norm": 1.507038950920105,
+      "epoch": 1.3618677042801557,
+      "grad_norm": 2.040506601333618,
       "learning_rate": 6.833333333333334e-06,
-      "loss": 0.0494,
+      "loss": 0.1548,
       "step": 1925
     },
     {
-      "epoch": 3.9117352056168504,
-      "grad_norm": 1.2737189531326294,
+      "epoch": 1.3795542978422355,
+      "grad_norm": 1.8651851415634155,
       "learning_rate": 6.777777777777779e-06,
-      "loss": 0.052,
+      "loss": 0.161,
       "step": 1950
     },
     {
-      "epoch": 3.961885656970913,
-      "grad_norm": 1.8823206424713135,
+      "epoch": 1.3972408914043155,
+      "grad_norm": 1.9047472476959229,
       "learning_rate": 6.7222222222222235e-06,
-      "loss": 0.053,
+      "loss": 0.1651,
       "step": 1975
     },
     {
-      "epoch": 4.012036108324975,
-      "grad_norm": 0.7559254765510559,
+      "epoch": 1.4149274849663955,
+      "grad_norm": 1.6781543493270874,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0472,
+      "loss": 0.1572,
       "step": 2000
     },
     {
-      "epoch": 4.012036108324975,
-      "eval_loss": 0.33296459913253784,
-      "eval_runtime": 4527.5772,
-      "eval_samples_per_second": 2.377,
-      "eval_steps_per_second": 0.149,
-      "eval_wer": 0.20142963310646952,
+      "epoch": 1.4149274849663955,
+      "eval_loss": 0.27680015563964844,
+      "eval_runtime": 4951.943,
+      "eval_samples_per_second": 2.173,
+      "eval_steps_per_second": 0.136,
+      "eval_wer": 0.18979308456543847,
       "step": 2000
     },
     {
-      "epoch": 4.062186559679037,
-      "grad_norm": 0.8223534822463989,
+      "epoch": 1.4326140785284753,
+      "grad_norm": 1.9535053968429565,
       "learning_rate": 6.6111111111111115e-06,
-      "loss": 0.0218,
+      "loss": 0.1586,
       "step": 2025
     },
     {
-      "epoch": 4.112337011033099,
-      "grad_norm": 0.7893447279930115,
+      "epoch": 1.4503006720905554,
+      "grad_norm": 1.7220150232315063,
       "learning_rate": 6.555555555555556e-06,
-      "loss": 0.0201,
+      "loss": 0.1634,
       "step": 2050
     },
     {
-      "epoch": 4.162487462387162,
-      "grad_norm": 0.657762348651886,
+      "epoch": 1.4679872656526354,
+      "grad_norm": 2.427858352661133,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0203,
+      "loss": 0.1659,
       "step": 2075
     },
     {
-      "epoch": 4.212637913741224,
-      "grad_norm": 0.9496999979019165,
+      "epoch": 1.4856738592147152,
+      "grad_norm": 1.9716796875,
       "learning_rate": 6.444444444444445e-06,
-      "loss": 0.018,
+      "loss": 0.1511,
       "step": 2100
     },
     {
-      "epoch": 4.262788365095286,
-      "grad_norm": 1.3166756629943848,
+      "epoch": 1.5033604527767952,
+      "grad_norm": 1.8595249652862549,
       "learning_rate": 6.3888888888888885e-06,
-      "loss": 0.0203,
+      "loss": 0.1556,
       "step": 2125
     },
     {
-      "epoch": 4.312938816449348,
-      "grad_norm": 0.8114544749259949,
+      "epoch": 1.5210470463388752,
+      "grad_norm": 2.3424742221832275,
       "learning_rate": 6.333333333333333e-06,
-      "loss": 0.022,
+      "loss": 0.1717,
       "step": 2150
     },
     {
-      "epoch": 4.36308926780341,
-      "grad_norm": 1.0165122747421265,
+      "epoch": 1.538733639900955,
+      "grad_norm": 2.172686815261841,
       "learning_rate": 6.277777777777778e-06,
-      "loss": 0.0188,
+      "loss": 0.1586,
       "step": 2175
     },
     {
-      "epoch": 4.413239719157472,
-      "grad_norm": 0.8669913411140442,
+      "epoch": 1.556420233463035,
+      "grad_norm": 1.7644214630126953,
       "learning_rate": 6.222222222222223e-06,
-      "loss": 0.0204,
+      "loss": 0.1521,
       "step": 2200
     },
     {
-      "epoch": 4.463390170511534,
-      "grad_norm": 1.483972430229187,
+      "epoch": 1.574106827025115,
+      "grad_norm": 2.1562883853912354,
       "learning_rate": 6.166666666666667e-06,
-      "loss": 0.0199,
+      "loss": 0.1578,
       "step": 2225
     },
     {
-      "epoch": 4.513540621865597,
-      "grad_norm": 1.1734397411346436,
+      "epoch": 1.5917934205871949,
+      "grad_norm": 2.087270736694336,
       "learning_rate": 6.111111111111112e-06,
-      "loss": 0.0194,
+      "loss": 0.1556,
       "step": 2250
     },
     {
-      "epoch": 4.563691073219659,
-      "grad_norm": 0.5651412606239319,
+      "epoch": 1.6094800141492749,
+      "grad_norm": 2.0405049324035645,
       "learning_rate": 6.055555555555555e-06,
-      "loss": 0.0219,
+      "loss": 0.1654,
       "step": 2275
     },
     {
-      "epoch": 4.613841524573721,
-      "grad_norm": 0.9663107395172119,
+      "epoch": 1.627166607711355,
+      "grad_norm": 1.6984881162643433,
       "learning_rate": 6e-06,
-      "loss": 0.0215,
+      "loss": 0.1573,
       "step": 2300
     },
     {
-      "epoch": 4.663991975927783,
-      "grad_norm": 0.6868607401847839,
+      "epoch": 1.6448532012734347,
+      "grad_norm": 2.290949583053589,
       "learning_rate": 5.944444444444445e-06,
-      "loss": 0.0223,
+      "loss": 0.1655,
       "step": 2325
     },
     {
-      "epoch": 4.714142427281845,
-      "grad_norm": 1.138351321220398,
+      "epoch": 1.6625397948355147,
+      "grad_norm": 1.9502941370010376,
       "learning_rate": 5.88888888888889e-06,
-      "loss": 0.0215,
+      "loss": 0.1462,
       "step": 2350
     },
     {
-      "epoch": 4.764292878635907,
-      "grad_norm": 0.7841187119483948,
+      "epoch": 1.6802263883975948,
+      "grad_norm": 1.9741941690444946,
       "learning_rate": 5.833333333333334e-06,
-      "loss": 0.0207,
+      "loss": 0.1636,
       "step": 2375
     },
     {
-      "epoch": 4.81444332998997,
-      "grad_norm": 0.6683496236801147,
+      "epoch": 1.6979129819596745,
+      "grad_norm": 1.9639830589294434,
       "learning_rate": 5.777777777777778e-06,
-      "loss": 0.0201,
+      "loss": 0.155,
       "step": 2400
     },
     {
-      "epoch": 4.864593781344032,
-      "grad_norm": 1.2674401998519897,
+      "epoch": 1.7155995755217543,
+      "grad_norm": 1.6349612474441528,
       "learning_rate": 5.722222222222222e-06,
-      "loss": 0.0196,
+      "loss": 0.1659,
       "step": 2425
     },
     {
-      "epoch": 4.914744232698094,
-      "grad_norm": 0.8776654601097107,
+      "epoch": 1.7332861690838346,
+      "grad_norm": 1.8674352169036865,
       "learning_rate": 5.666666666666667e-06,
-      "loss": 0.0214,
+      "loss": 0.152,
       "step": 2450
     },
     {
-      "epoch": 4.964894684052156,
-      "grad_norm": 0.8897686004638672,
+      "epoch": 1.7509727626459144,
+      "grad_norm": 2.1415135860443115,
       "learning_rate": 5.611111111111112e-06,
-      "loss": 0.0201,
+      "loss": 0.1699,
       "step": 2475
     },
     {
-      "epoch": 5.015045135406218,
-      "grad_norm": 0.6467758417129517,
+      "epoch": 1.7686593562079942,
+      "grad_norm": 1.5318387746810913,
       "learning_rate": 5.555555555555557e-06,
-      "loss": 0.0176,
+      "loss": 0.1592,
       "step": 2500
     },
     {
-      "epoch": 5.0651955867602805,
-      "grad_norm": 0.8919677734375,
+      "epoch": 1.7863459497700744,
+      "grad_norm": 1.758480191230774,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.0099,
+      "loss": 0.1537,
       "step": 2525
     },
     {
-      "epoch": 5.115346038114343,
-      "grad_norm": 0.23880383372306824,
+      "epoch": 1.8040325433321542,
+      "grad_norm": 1.815434455871582,
       "learning_rate": 5.444444444444445e-06,
-      "loss": 0.0089,
+      "loss": 0.1533,
       "step": 2550
     },
     {
-      "epoch": 5.1654964894684054,
-      "grad_norm": 0.7303509712219238,
+      "epoch": 1.821719136894234,
+      "grad_norm": 1.5665532350540161,
       "learning_rate": 5.388888888888889e-06,
-      "loss": 0.0093,
+      "loss": 0.1601,
       "step": 2575
     },
     {
-      "epoch": 5.2156469408224675,
-      "grad_norm": 0.8609201312065125,
+      "epoch": 1.839405730456314,
+      "grad_norm": 2.0574448108673096,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 0.0083,
+      "loss": 0.1548,
       "step": 2600
     },
     {
-      "epoch": 5.2657973921765295,
-      "grad_norm": 0.4490683078765869,
+      "epoch": 1.857092324018394,
+      "grad_norm": 1.7183318138122559,
       "learning_rate": 5.2777777777777785e-06,
-      "loss": 0.009,
+      "loss": 0.1477,
       "step": 2625
     },
     {
-      "epoch": 5.315947843530592,
-      "grad_norm": 0.5905369520187378,
+      "epoch": 1.8747789175804739,
+      "grad_norm": 1.7662074565887451,
       "learning_rate": 5.2222222222222226e-06,
-      "loss": 0.009,
+      "loss": 0.1486,
       "step": 2650
     },
     {
-      "epoch": 5.366098294884654,
-      "grad_norm": 0.18808700144290924,
+      "epoch": 1.892465511142554,
+      "grad_norm": 1.840253472328186,
       "learning_rate": 5.1666666666666675e-06,
-      "loss": 0.0075,
+      "loss": 0.1516,
       "step": 2675
     },
     {
-      "epoch": 5.4162487462387165,
-      "grad_norm": 0.8524003624916077,
+      "epoch": 1.910152104704634,
+      "grad_norm": 1.9791252613067627,
       "learning_rate": 5.1111111111111115e-06,
-      "loss": 0.0084,
+      "loss": 0.1431,
       "step": 2700
     },
     {
-      "epoch": 5.466399197592779,
-      "grad_norm": 0.31983518600463867,
+      "epoch": 1.9278386982667137,
+      "grad_norm": 1.5719287395477295,
       "learning_rate": 5.0555555555555555e-06,
-      "loss": 0.0081,
+      "loss": 0.1539,
       "step": 2725
     },
     {
-      "epoch": 5.516549648946841,
-      "grad_norm": 0.40100687742233276,
+      "epoch": 1.9455252918287937,
+      "grad_norm": 2.328838348388672,
       "learning_rate": 5e-06,
-      "loss": 0.0077,
+      "loss": 0.1502,
       "step": 2750
     },
     {
-      "epoch": 5.566700100300903,
-      "grad_norm": 1.5288898944854736,
+      "epoch": 1.9632118853908738,
+      "grad_norm": 1.7131285667419434,
       "learning_rate": 4.944444444444445e-06,
-      "loss": 0.0077,
+      "loss": 0.1469,
       "step": 2775
     },
     {
-      "epoch": 5.616850551654965,
-      "grad_norm": 0.15371347963809967,
+      "epoch": 1.9808984789529536,
+      "grad_norm": 1.751104712486267,
       "learning_rate": 4.888888888888889e-06,
-      "loss": 0.0068,
+      "loss": 0.151,
       "step": 2800
     },
     {
-      "epoch": 5.667001003009027,
-      "grad_norm": 0.547875702381134,
+      "epoch": 1.9985850725150336,
+      "grad_norm": 1.8015530109405518,
       "learning_rate": 4.833333333333333e-06,
-      "loss": 0.0078,
+      "loss": 0.1446,
       "step": 2825
     },
     {
-      "epoch": 5.717151454363089,
-      "grad_norm": 0.3743741512298584,
+      "epoch": 2.0162716660771136,
+      "grad_norm": 1.4693965911865234,
       "learning_rate": 4.777777777777778e-06,
-      "loss": 0.0105,
+      "loss": 0.0913,
       "step": 2850
     },
     {
-      "epoch": 5.767301905717152,
-      "grad_norm": 1.20223867893219,
+      "epoch": 2.0339582596391934,
+      "grad_norm": 1.5922882556915283,
       "learning_rate": 4.722222222222222e-06,
-      "loss": 0.0084,
+      "loss": 0.0824,
       "step": 2875
     },
     {
-      "epoch": 5.817452357071214,
-      "grad_norm": 0.4010525047779083,
+      "epoch": 2.0516448532012737,
+      "grad_norm": 1.0818850994110107,
       "learning_rate": 4.666666666666667e-06,
-      "loss": 0.0074,
+      "loss": 0.08,
       "step": 2900
     },
     {
-      "epoch": 5.867602808425276,
-      "grad_norm": 0.2383079081773758,
+      "epoch": 2.0693314467633535,
+      "grad_norm": 1.5331037044525146,
       "learning_rate": 4.611111111111112e-06,
-      "loss": 0.0097,
+      "loss": 0.0761,
       "step": 2925
     },
     {
-      "epoch": 5.917753259779338,
-      "grad_norm": 0.9694074988365173,
+      "epoch": 2.0870180403254333,
+      "grad_norm": 1.4420839548110962,
       "learning_rate": 4.555555555555556e-06,
-      "loss": 0.0082,
+      "loss": 0.072,
       "step": 2950
     },
     {
-      "epoch": 5.9679037111334,
-      "grad_norm": 0.4873412549495697,
+      "epoch": 2.104704633887513,
+      "grad_norm": 1.4210786819458008,
       "learning_rate": 4.5e-06,
-      "loss": 0.0094,
+      "loss": 0.073,
       "step": 2975
     },
     {
-      "epoch": 6.018054162487463,
-      "grad_norm": 0.35399332642555237,
+      "epoch": 2.1223912274495933,
+      "grad_norm": 1.2918660640716553,
       "learning_rate": 4.444444444444444e-06,
-      "loss": 0.0074,
+      "loss": 0.0714,
       "step": 3000
     },
     {
-      "epoch": 6.018054162487463,
-      "eval_loss": 0.3955759108066559,
-      "eval_runtime": 4515.9024,
-      "eval_samples_per_second": 2.383,
-      "eval_steps_per_second": 0.149,
-      "eval_wer": 0.19721505123992802,
+      "epoch": 2.1223912274495933,
+      "eval_loss": 0.27894869446754456,
+      "eval_runtime": 4698.3805,
+      "eval_samples_per_second": 2.291,
+      "eval_steps_per_second": 0.143,
+      "eval_wer": 0.18066963936478134,
       "step": 3000
     },
     {
-      "epoch": 6.068204613841525,
-      "grad_norm": 0.32993289828300476,
+      "epoch": 2.140077821011673,
+      "grad_norm": 1.2981915473937988,
       "learning_rate": 4.388888888888889e-06,
-      "loss": 0.005,
+      "loss": 0.0705,
       "step": 3025
     },
     {
-      "epoch": 6.118355065195587,
-      "grad_norm": 2.4931156635284424,
+      "epoch": 2.157764414573753,
+      "grad_norm": 1.5843642950057983,
       "learning_rate": 4.333333333333334e-06,
-      "loss": 0.0047,
+      "loss": 0.0801,
       "step": 3050
     },
     {
-      "epoch": 6.168505516549649,
-      "grad_norm": 0.09684254229068756,
+      "epoch": 2.175451008135833,
+      "grad_norm": 1.4696415662765503,
       "learning_rate": 4.277777777777778e-06,
-      "loss": 0.0039,
+      "loss": 0.0828,
       "step": 3075
     },
     {
-      "epoch": 6.218655967903711,
-      "grad_norm": 0.1377803087234497,
+      "epoch": 2.193137601697913,
+      "grad_norm": 1.4634337425231934,
       "learning_rate": 4.222222222222223e-06,
-      "loss": 0.0041,
+      "loss": 0.0744,
       "step": 3100
     },
     {
-      "epoch": 6.268806419257773,
-      "grad_norm": 0.12329312413930893,
+      "epoch": 2.2108241952599927,
+      "grad_norm": 1.241773247718811,
       "learning_rate": 4.166666666666667e-06,
-      "loss": 0.0036,
+      "loss": 0.0735,
       "step": 3125
     },
     {
-      "epoch": 6.318956870611835,
-      "grad_norm": 0.39106011390686035,
+      "epoch": 2.228510788822073,
+      "grad_norm": 1.7622851133346558,
       "learning_rate": 4.111111111111111e-06,
-      "loss": 0.0038,
+      "loss": 0.0755,
       "step": 3150
     },
     {
-      "epoch": 6.369107321965898,
-      "grad_norm": 0.21228019893169403,
+      "epoch": 2.246197382384153,
+      "grad_norm": 1.7847949266433716,
       "learning_rate": 4.055555555555556e-06,
-      "loss": 0.0034,
+      "loss": 0.0697,
       "step": 3175
     },
     {
-      "epoch": 6.41925777331996,
-      "grad_norm": 0.13951948285102844,
+      "epoch": 2.2638839759462326,
+      "grad_norm": 1.7188225984573364,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0035,
+      "loss": 0.0783,
       "step": 3200
     },
     {
-      "epoch": 6.469408224674022,
-      "grad_norm": 0.15315498411655426,
+      "epoch": 2.281570569508313,
+      "grad_norm": 1.4745049476623535,
       "learning_rate": 3.944444444444445e-06,
-      "loss": 0.0045,
+      "loss": 0.0782,
       "step": 3225
     },
     {
-      "epoch": 6.519558676028084,
-      "grad_norm": 0.09918154031038284,
+      "epoch": 2.2992571630703926,
+      "grad_norm": 1.488161563873291,
       "learning_rate": 3.88888888888889e-06,
-      "loss": 0.0037,
+      "loss": 0.081,
       "step": 3250
     },
     {
-      "epoch": 6.569709127382146,
-      "grad_norm": 0.128856360912323,
+      "epoch": 2.3169437566324724,
+      "grad_norm": 1.4803715944290161,
       "learning_rate": 3.833333333333334e-06,
-      "loss": 0.0034,
+      "loss": 0.0734,
       "step": 3275
     },
     {
-      "epoch": 6.619859578736209,
-      "grad_norm": 0.35564741492271423,
+      "epoch": 2.3346303501945527,
+      "grad_norm": 1.7518547773361206,
       "learning_rate": 3.777777777777778e-06,
-      "loss": 0.0035,
+      "loss": 0.0791,
       "step": 3300
     },
     {
-      "epoch": 6.670010030090271,
-      "grad_norm": 0.09503413736820221,
+      "epoch": 2.3523169437566325,
+      "grad_norm": 1.9586306810379028,
       "learning_rate": 3.7222222222222225e-06,
-      "loss": 0.0047,
+      "loss": 0.0814,
       "step": 3325
     },
     {
-      "epoch": 6.720160481444333,
-      "grad_norm": 0.2272273600101471,
+      "epoch": 2.3700035373187123,
+      "grad_norm": 1.8565925359725952,
       "learning_rate": 3.6666666666666666e-06,
-      "loss": 0.004,
+      "loss": 0.0728,
       "step": 3350
     },
     {
-      "epoch": 6.770310932798395,
-      "grad_norm": 0.31332728266716003,
+      "epoch": 2.3876901308807925,
+      "grad_norm": 1.5781797170639038,
       "learning_rate": 3.6111111111111115e-06,
-      "loss": 0.0038,
+      "loss": 0.0826,
       "step": 3375
     },
     {
-      "epoch": 6.820461384152457,
-      "grad_norm": 0.20381228625774384,
+      "epoch": 2.4053767244428723,
+      "grad_norm": 1.5568901300430298,
       "learning_rate": 3.555555555555556e-06,
-      "loss": 0.0044,
+      "loss": 0.0802,
       "step": 3400
     },
     {
-      "epoch": 6.870611835506519,
-      "grad_norm": 0.38565367460250854,
+      "epoch": 2.423063318004952,
+      "grad_norm": 1.2662155628204346,
       "learning_rate": 3.5e-06,
-      "loss": 0.0052,
+      "loss": 0.074,
       "step": 3425
     },
     {
-      "epoch": 6.920762286860581,
-      "grad_norm": 0.1613508015871048,
+      "epoch": 2.4407499115670324,
+      "grad_norm": 1.559577226638794,
       "learning_rate": 3.444444444444445e-06,
-      "loss": 0.0037,
+      "loss": 0.0758,
       "step": 3450
     },
     {
-      "epoch": 6.970912738214644,
-      "grad_norm": 0.23183399438858032,
+      "epoch": 2.458436505129112,
+      "grad_norm": 1.5125735998153687,
       "learning_rate": 3.3888888888888893e-06,
-      "loss": 0.0042,
+      "loss": 0.0756,
       "step": 3475
     },
     {
-      "epoch": 7.021063189568706,
-      "grad_norm": 0.09113147854804993,
+      "epoch": 2.476123098691192,
+      "grad_norm": 1.4088867902755737,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.0028,
+      "loss": 0.0713,
       "step": 3500
     },
     {
-      "epoch": 7.071213640922768,
-      "grad_norm": 0.12553289532661438,
+      "epoch": 2.493809692253272,
+      "grad_norm": 1.6018534898757935,
       "learning_rate": 3.277777777777778e-06,
-      "loss": 0.0026,
+      "loss": 0.0656,
       "step": 3525
     },
     {
-      "epoch": 7.12136409227683,
-      "grad_norm": 0.0897342711687088,
+      "epoch": 2.511496285815352,
+      "grad_norm": 1.5624502897262573,
       "learning_rate": 3.2222222222222227e-06,
-      "loss": 0.002,
+      "loss": 0.0802,
       "step": 3550
     },
     {
-      "epoch": 7.1715145436308925,
-      "grad_norm": 0.07827088236808777,
+      "epoch": 2.529182879377432,
+      "grad_norm": 1.4694902896881104,
       "learning_rate": 3.1666666666666667e-06,
-      "loss": 0.0024,
+      "loss": 0.0713,
       "step": 3575
     },
     {
-      "epoch": 7.2216649949849545,
-      "grad_norm": 0.055039647966623306,
+      "epoch": 2.5468694729395116,
+      "grad_norm": 1.6186368465423584,
       "learning_rate": 3.1111111111111116e-06,
-      "loss": 0.0024,
+      "loss": 0.0864,
       "step": 3600
     },
     {
-      "epoch": 7.271815446339017,
-      "grad_norm": 0.06994307041168213,
+      "epoch": 2.564556066501592,
+      "grad_norm": 1.6998704671859741,
       "learning_rate": 3.055555555555556e-06,
-      "loss": 0.0021,
+      "loss": 0.0721,
       "step": 3625
     },
     {
-      "epoch": 7.3219658976930795,
-      "grad_norm": 0.10969334840774536,
+      "epoch": 2.5822426600636716,
+      "grad_norm": 1.7324111461639404,
       "learning_rate": 3e-06,
-      "loss": 0.0023,
+      "loss": 0.0747,
       "step": 3650
     },
     {
-      "epoch": 7.3721163490471415,
-      "grad_norm": 0.08186643570661545,
+      "epoch": 2.599929253625752,
+      "grad_norm": 1.2067221403121948,
       "learning_rate": 2.944444444444445e-06,
-      "loss": 0.0022,
+      "loss": 0.0738,
       "step": 3675
     },
     {
-      "epoch": 7.422266800401204,
-      "grad_norm": 0.08872943371534348,
+      "epoch": 2.6176158471878317,
+      "grad_norm": 1.8782434463500977,
       "learning_rate": 2.888888888888889e-06,
-      "loss": 0.0027,
+      "loss": 0.0699,
       "step": 3700
     },
     {
-      "epoch": 7.472417251755266,
-      "grad_norm": 0.20957154035568237,
+      "epoch": 2.6353024407499115,
+      "grad_norm": 1.240626335144043,
       "learning_rate": 2.8333333333333335e-06,
-      "loss": 0.0027,
+      "loss": 0.0747,
       "step": 3725
     },
     {
-      "epoch": 7.522567703109328,
-      "grad_norm": 0.09654885530471802,
+      "epoch": 2.6529890343119913,
+      "grad_norm": 1.2365247011184692,
       "learning_rate": 2.7777777777777783e-06,
-      "loss": 0.0021,
+      "loss": 0.0734,
       "step": 3750
     },
     {
-      "epoch": 7.57271815446339,
-      "grad_norm": 0.10487361997365952,
+      "epoch": 2.6706756278740715,
+      "grad_norm": 1.5169546604156494,
       "learning_rate": 2.7222222222222224e-06,
-      "loss": 0.0041,
+      "loss": 0.0734,
       "step": 3775
     },
     {
-      "epoch": 7.622868605817453,
-      "grad_norm": 0.06899241358041763,
+      "epoch": 2.6883622214361513,
+      "grad_norm": 1.3478069305419922,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 0.0025,
+      "loss": 0.0757,
       "step": 3800
     },
     {
-      "epoch": 7.673019057171515,
-      "grad_norm": 0.06502360850572586,
+      "epoch": 2.7060488149982316,
+      "grad_norm": 1.938991665840149,
       "learning_rate": 2.6111111111111113e-06,
-      "loss": 0.0025,
+      "loss": 0.0774,
       "step": 3825
     },
     {
-      "epoch": 7.723169508525577,
-      "grad_norm": 0.05213673412799835,
+      "epoch": 2.7237354085603114,
+      "grad_norm": 1.7492825984954834,
       "learning_rate": 2.5555555555555557e-06,
-      "loss": 0.0026,
+      "loss": 0.0679,
       "step": 3850
     },
     {
-      "epoch": 7.773319959879639,
-      "grad_norm": 0.11386407911777496,
+      "epoch": 2.741422002122391,
+      "grad_norm": 1.7618768215179443,
       "learning_rate": 2.5e-06,
-      "loss": 0.0023,
+      "loss": 0.0702,
       "step": 3875
     },
     {
-      "epoch": 7.823470411233701,
-      "grad_norm": 0.08614542335271835,
+      "epoch": 2.759108595684471,
+      "grad_norm": 1.836369276046753,
       "learning_rate": 2.4444444444444447e-06,
-      "loss": 0.002,
+      "loss": 0.0707,
       "step": 3900
     },
     {
-      "epoch": 7.873620862587764,
-      "grad_norm": 0.12035714089870453,
+      "epoch": 2.776795189246551,
+      "grad_norm": 1.0155326128005981,
       "learning_rate": 2.388888888888889e-06,
-      "loss": 0.0021,
+      "loss": 0.0692,
       "step": 3925
     },
     {
-      "epoch": 7.923771313941826,
-      "grad_norm": 0.0683644562959671,
+      "epoch": 2.794481782808631,
+      "grad_norm": 1.5357517004013062,
       "learning_rate": 2.3333333333333336e-06,
-      "loss": 0.0023,
+      "loss": 0.064,
       "step": 3950
     },
     {
-      "epoch": 7.973921765295888,
-      "grad_norm": 0.059945497661828995,
+      "epoch": 2.8121683763707113,
+      "grad_norm": 1.413317322731018,
       "learning_rate": 2.277777777777778e-06,
-      "loss": 0.002,
+      "loss": 0.0761,
       "step": 3975
     },
     {
-      "epoch": 8.02407221664995,
-      "grad_norm": 0.050406791269779205,
+      "epoch": 2.829854969932791,
+      "grad_norm": 1.81116783618927,
       "learning_rate": 2.222222222222222e-06,
-      "loss": 0.0019,
+      "loss": 0.0772,
       "step": 4000
     },
     {
-      "epoch": 8.02407221664995,
-      "eval_loss": 0.4262579679489136,
-      "eval_runtime": 4522.1238,
-      "eval_samples_per_second": 2.38,
-      "eval_steps_per_second": 0.149,
-      "eval_wer": 0.1952691074082766,
+      "epoch": 2.829854969932791,
+      "eval_loss": 0.27590474486351013,
+      "eval_runtime": 4660.1057,
+      "eval_samples_per_second": 2.31,
+      "eval_steps_per_second": 0.144,
+      "eval_wer": 0.18098255495580068,
       "step": 4000
     },
     {
-      "epoch": 8.074222668004012,
-      "grad_norm": 0.04868149757385254,
+      "epoch": 2.847541563494871,
+      "grad_norm": 1.819291591644287,
       "learning_rate": 2.166666666666667e-06,
-      "loss": 0.0016,
+      "loss": 0.0739,
       "step": 4025
     },
     {
-      "epoch": 8.124373119358074,
-      "grad_norm": 0.0494823083281517,
+      "epoch": 2.8652281570569507,
+      "grad_norm": 1.4449914693832397,
       "learning_rate": 2.1111111111111114e-06,
-      "loss": 0.0018,
+      "loss": 0.0672,
       "step": 4050
     },
     {
-      "epoch": 8.174523570712136,
-      "grad_norm": 0.05864572152495384,
+      "epoch": 2.882914750619031,
+      "grad_norm": 1.617101788520813,
       "learning_rate": 2.0555555555555555e-06,
-      "loss": 0.0017,
+      "loss": 0.072,
       "step": 4075
     },
     {
-      "epoch": 8.224674022066198,
-      "grad_norm": 0.05070830136537552,
+      "epoch": 2.9006013441811107,
+      "grad_norm": 1.7751729488372803,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0016,
+      "loss": 0.075,
       "step": 4100
     },
     {
-      "epoch": 8.27482447342026,
-      "grad_norm": 0.06109990179538727,
+      "epoch": 2.9182879377431905,
+      "grad_norm": 1.939385175704956,
       "learning_rate": 1.944444444444445e-06,
-      "loss": 0.0017,
+      "loss": 0.0746,
       "step": 4125
     },
     {
-      "epoch": 8.324974924774324,
-      "grad_norm": 0.045594800263643265,
+      "epoch": 2.9359745313052708,
+      "grad_norm": 1.2767341136932373,
       "learning_rate": 1.888888888888889e-06,
-      "loss": 0.0017,
+      "loss": 0.065,
       "step": 4150
     },
     {
-      "epoch": 8.375125376128386,
-      "grad_norm": 0.049554213881492615,
+      "epoch": 2.9536611248673506,
+      "grad_norm": 1.384352445602417,
       "learning_rate": 1.8333333333333333e-06,
-      "loss": 0.0021,
+      "loss": 0.0729,
       "step": 4175
     },
     {
-      "epoch": 8.425275827482448,
-      "grad_norm": 0.11533299088478088,
+      "epoch": 2.9713477184294304,
+      "grad_norm": 1.707626223564148,
       "learning_rate": 1.777777777777778e-06,
-      "loss": 0.0015,
+      "loss": 0.0699,
       "step": 4200
     },
     {
-      "epoch": 8.47542627883651,
-      "grad_norm": 0.043509747833013535,
+      "epoch": 2.9890343119915106,
+      "grad_norm": 1.5926012992858887,
       "learning_rate": 1.7222222222222224e-06,
-      "loss": 0.0018,
+      "loss": 0.0653,
       "step": 4225
     },
     {
-      "epoch": 8.525576730190572,
-      "grad_norm": 0.14743666350841522,
+      "epoch": 3.0067209055535904,
+      "grad_norm": 0.7629415988922119,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.002,
+      "loss": 0.0602,
       "step": 4250
     },
     {
-      "epoch": 8.575727181544634,
-      "grad_norm": 0.08023034036159515,
+      "epoch": 3.02440749911567,
+      "grad_norm": 1.042358636856079,
       "learning_rate": 1.6111111111111113e-06,
-      "loss": 0.0017,
+      "loss": 0.0343,
       "step": 4275
     },
     {
-      "epoch": 8.625877632898696,
-      "grad_norm": 0.1176542118191719,
+      "epoch": 3.0420940926777504,
+      "grad_norm": 1.4969948530197144,
       "learning_rate": 1.5555555555555558e-06,
-      "loss": 0.0019,
+      "loss": 0.0375,
       "step": 4300
     },
     {
-      "epoch": 8.676028084252758,
-      "grad_norm": 0.05018528178334236,
+      "epoch": 3.0597806862398302,
+      "grad_norm": 2.0350911617279053,
       "learning_rate": 1.5e-06,
-      "loss": 0.0018,
+      "loss": 0.0343,
       "step": 4325
     },
     {
-      "epoch": 8.72617853560682,
-      "grad_norm": 0.051610466092824936,
+      "epoch": 3.07746727980191,
+      "grad_norm": 1.1933727264404297,
       "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.0016,
+      "loss": 0.0419,
       "step": 4350
     },
     {
-      "epoch": 8.776328986960882,
-      "grad_norm": 0.04370126128196716,
+      "epoch": 3.0951538733639903,
+      "grad_norm": 1.2081432342529297,
       "learning_rate": 1.3888888888888892e-06,
-      "loss": 0.0016,
+      "loss": 0.0349,
       "step": 4375
     },
     {
-      "epoch": 8.826479438314944,
-      "grad_norm": 0.06507682055234909,
+      "epoch": 3.11284046692607,
+      "grad_norm": 1.3141449689865112,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.0015,
+      "loss": 0.0357,
       "step": 4400
     },
     {
-      "epoch": 8.876629889669006,
-      "grad_norm": 0.06819342076778412,
+      "epoch": 3.13052706048815,
+      "grad_norm": 1.156983494758606,
       "learning_rate": 1.2777777777777779e-06,
-      "loss": 0.0019,
+      "loss": 0.0313,
       "step": 4425
     },
     {
-      "epoch": 8.926780341023068,
-      "grad_norm": 0.06309027969837189,
+      "epoch": 3.14821365405023,
+      "grad_norm": 0.8983932137489319,
       "learning_rate": 1.2222222222222223e-06,
-      "loss": 0.0017,
+      "loss": 0.0324,
       "step": 4450
     },
     {
-      "epoch": 8.976930792377132,
-      "grad_norm": 0.08104189485311508,
+      "epoch": 3.16590024761231,
+      "grad_norm": 0.7461453676223755,
       "learning_rate": 1.1666666666666668e-06,
-      "loss": 0.0015,
+      "loss": 0.0414,
       "step": 4475
     },
     {
-      "epoch": 9.027081243731194,
-      "grad_norm": 0.04061448946595192,
+      "epoch": 3.1835868411743897,
+      "grad_norm": 1.092787265777588,
       "learning_rate": 1.111111111111111e-06,
-      "loss": 0.0015,
+      "loss": 0.0363,
       "step": 4500
     },
     {
-      "epoch": 9.077231695085256,
-      "grad_norm": 0.04562969505786896,
+      "epoch": 3.2012734347364695,
+      "grad_norm": 1.2066584825515747,
       "learning_rate": 1.0555555555555557e-06,
-      "loss": 0.0018,
+      "loss": 0.0361,
       "step": 4525
     },
     {
-      "epoch": 9.127382146439318,
-      "grad_norm": 0.050402842462062836,
+      "epoch": 3.2189600282985498,
+      "grad_norm": 0.7660400867462158,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0015,
+      "loss": 0.0338,
       "step": 4550
     },
     {
-      "epoch": 9.17753259779338,
-      "grad_norm": 0.5361291170120239,
+      "epoch": 3.2366466218606296,
+      "grad_norm": 0.8779363632202148,
       "learning_rate": 9.444444444444445e-07,
-      "loss": 0.0019,
+      "loss": 0.0416,
       "step": 4575
     },
     {
-      "epoch": 9.227683049147442,
-      "grad_norm": 0.0427754744887352,
+      "epoch": 3.25433321542271,
+      "grad_norm": 1.0976966619491577,
       "learning_rate": 8.88888888888889e-07,
-      "loss": 0.0015,
+      "loss": 0.0381,
       "step": 4600
     },
     {
-      "epoch": 9.277833500501504,
-      "grad_norm": 0.04894596338272095,
+      "epoch": 3.2720198089847896,
+      "grad_norm": 0.977162778377533,
       "learning_rate": 8.333333333333333e-07,
-      "loss": 0.0015,
+      "loss": 0.0318,
       "step": 4625
     },
     {
-      "epoch": 9.327983951855567,
-      "grad_norm": 0.04485371708869934,
+      "epoch": 3.2897064025468694,
+      "grad_norm": 1.5296571254730225,
       "learning_rate": 7.777777777777779e-07,
-      "loss": 0.0014,
+      "loss": 0.037,
       "step": 4650
     },
     {
-      "epoch": 9.378134403209629,
-      "grad_norm": 0.04477512091398239,
+      "epoch": 3.307392996108949,
+      "grad_norm": 1.0584012269973755,
       "learning_rate": 7.222222222222222e-07,
-      "loss": 0.0015,
+      "loss": 0.0369,
       "step": 4675
     },
     {
-      "epoch": 9.42828485456369,
-      "grad_norm": 0.04668520390987396,
+      "epoch": 3.3250795896710295,
+      "grad_norm": 0.8479712009429932,
       "learning_rate": 6.666666666666667e-07,
-      "loss": 0.0015,
+      "loss": 0.0343,
       "step": 4700
     },
     {
-      "epoch": 9.478435305917753,
-      "grad_norm": 0.05096197500824928,
+      "epoch": 3.3427661832331093,
+      "grad_norm": 1.176857352256775,
       "learning_rate": 6.111111111111112e-07,
-      "loss": 0.0017,
+      "loss": 0.0349,
       "step": 4725
     },
     {
-      "epoch": 9.528585757271815,
-      "grad_norm": 0.04469752684235573,
+      "epoch": 3.360452776795189,
+      "grad_norm": 1.1220146417617798,
       "learning_rate": 5.555555555555555e-07,
-      "loss": 0.0015,
+      "loss": 0.0307,
       "step": 4750
     },
     {
-      "epoch": 9.578736208625877,
-      "grad_norm": 0.050533875823020935,
+      "epoch": 3.3781393703572693,
+      "grad_norm": 0.9692522883415222,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0014,
+      "loss": 0.0345,
       "step": 4775
     },
     {
-      "epoch": 9.62888665997994,
-      "grad_norm": 0.04058687761425972,
+      "epoch": 3.395825963919349,
+      "grad_norm": 0.8425891995429993,
       "learning_rate": 4.444444444444445e-07,
-      "loss": 0.0015,
+      "loss": 0.0373,
       "step": 4800
     },
     {
-      "epoch": 9.679037111334003,
-      "grad_norm": 0.05510278418660164,
+      "epoch": 3.413512557481429,
+      "grad_norm": 1.2310644388198853,
       "learning_rate": 3.8888888888888895e-07,
-      "loss": 0.0014,
+      "loss": 0.032,
       "step": 4825
     },
     {
-      "epoch": 9.729187562688065,
-      "grad_norm": 0.05432584136724472,
+      "epoch": 3.431199151043509,
+      "grad_norm": 1.2488874197006226,
       "learning_rate": 3.3333333333333335e-07,
-      "loss": 0.0013,
+      "loss": 0.033,
       "step": 4850
     },
     {
-      "epoch": 9.779338014042127,
-      "grad_norm": 0.08223077654838562,
+      "epoch": 3.448885744605589,
+      "grad_norm": 0.8309725522994995,
       "learning_rate": 2.7777777777777776e-07,
-      "loss": 0.0014,
+      "loss": 0.035,
       "step": 4875
     },
     {
-      "epoch": 9.829488465396189,
-      "grad_norm": 0.051733825355768204,
+      "epoch": 3.4665723381676687,
+      "grad_norm": 0.8922958970069885,
       "learning_rate": 2.2222222222222224e-07,
-      "loss": 0.0014,
+      "loss": 0.0306,
       "step": 4900
     },
     {
-      "epoch": 9.87963891675025,
-      "grad_norm": 0.03811487555503845,
+      "epoch": 3.484258931729749,
+      "grad_norm": 0.7373726963996887,
       "learning_rate": 1.6666666666666668e-07,
-      "loss": 0.0013,
+      "loss": 0.0324,
       "step": 4925
     },
     {
-      "epoch": 9.929789368104313,
-      "grad_norm": 0.03971032425761223,
+      "epoch": 3.501945525291829,
+      "grad_norm": 1.3404922485351562,
       "learning_rate": 1.1111111111111112e-07,
-      "loss": 0.0013,
+      "loss": 0.0362,
       "step": 4950
     },
     {
-      "epoch": 9.979939819458375,
-      "grad_norm": 0.0468224436044693,
+      "epoch": 3.5196321188539086,
+      "grad_norm": 1.187138319015503,
       "learning_rate": 5.555555555555556e-08,
-      "loss": 0.0016,
+      "loss": 0.0373,
       "step": 4975
     },
     {
-      "epoch": 10.030090270812437,
-      "grad_norm": 0.04448791965842247,
+      "epoch": 3.537318712415989,
+      "grad_norm": 0.8413584232330322,
       "learning_rate": 0.0,
-      "loss": 0.0014,
+      "loss": 0.0337,
       "step": 5000
     },
     {
-      "epoch": 10.030090270812437,
-      "eval_loss": 0.4447694718837738,
-      "eval_runtime": 4528.3909,
-      "eval_samples_per_second": 2.377,
-      "eval_steps_per_second": 0.149,
-      "eval_wer": 0.1960611749980443,
+      "epoch": 3.537318712415989,
+      "eval_loss": 0.3035752773284912,
+      "eval_runtime": 4623.5314,
+      "eval_samples_per_second": 2.328,
+      "eval_steps_per_second": 0.146,
+      "eval_wer": 0.18264491903309082,
       "step": 5000
     },
     {
-      "epoch": 10.030090270812437,
+      "epoch": 3.537318712415989,
       "step": 5000,
-      "total_flos": 5.43090104303616e+20,
-      "train_loss": 0.0887994671344757,
-      "train_runtime": 50408.8653,
-      "train_samples_per_second": 3.174,
-      "train_steps_per_second": 0.099
+      "total_flos": 5.435079965953229e+20,
+      "train_loss": 0.17054095619916915,
+      "train_runtime": 54364.2112,
+      "train_samples_per_second": 2.943,
+      "train_steps_per_second": 0.092
     }
   ],
   "logging_steps": 25,
   "max_steps": 5000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 11,
+  "num_train_epochs": 4,
   "save_steps": 1000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -1480,7 +1480,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.43090104303616e+20,
+  "total_flos": 5.435079965953229e+20,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null