End of training

README.md

@@ -3,11 +3,24 @@ license: apache-2.0
 base_model: openai/whisper-large-v3
 tags:
 - generated_from_trainer
+datasets:
+- DewiBrynJones/oscar-cy-tts
 metrics:
 - wer
 model-index:
 - name: whisper-large-v3-ft-tts-cy
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: DewiBrynJones/oscar-cy-tts default
+      type: DewiBrynJones/oscar-cy-tts
+      args: default
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.10755268881722042
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -15,7 +28,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-large-v3-ft-tts-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/oscar-cy-tts default dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.1365
 - Wer: 0.1076

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 1.3137151865475565,
-    "eval_loss": 0.21081987023353577,
-    "eval_runtime": 675.2588,
-    "eval_samples": 1220,
-    "eval_samples_per_second": 1.807,
-    "eval_steps_per_second": 0.114,
-    "eval_wer": 0.1561639017527405,
+    "epoch": 1.6160310277957337,
+    "eval_loss": 0.1364615559577942,
+    "eval_runtime": 431.3495,
+    "eval_samples": 1000,
+    "eval_samples_per_second": 2.318,
+    "eval_steps_per_second": 0.146,
+    "eval_wer": 0.10755268881722042,
     "total_flos": 5.435725490631475e+20,
-    "train_loss": 0.2802161669254303,
-    "train_runtime": 36290.7707,
-    "train_samples": 121784,
-    "train_samples_per_second": 4.409,
-    "train_steps_per_second": 0.138
+    "train_loss": 0.17819489703178407,
+    "train_runtime": 33619.5535,
+    "train_samples": 99000,
+    "train_samples_per_second": 4.759,
+    "train_steps_per_second": 0.149
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 1.3137151865475565,
-    "eval_loss": 0.21081987023353577,
-    "eval_runtime": 675.2588,
-    "eval_samples": 1220,
-    "eval_samples_per_second": 1.807,
-    "eval_steps_per_second": 0.114,
-    "eval_wer": 0.1561639017527405
+    "epoch": 1.6160310277957337,
+    "eval_loss": 0.1364615559577942,
+    "eval_runtime": 431.3495,
+    "eval_samples": 1000,
+    "eval_samples_per_second": 2.318,
+    "eval_steps_per_second": 0.146,
+    "eval_wer": 0.10755268881722042
 }

runs/Aug27_07-38-23_ea28740c24c9/events.out.tfevents.1724776539.ea28740c24c9.2076.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98d39567b66faf1ef0141af636c765b831864a03a1f3a1ee5778fa7ac701fce6
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 1.3137151865475565,
+    "epoch": 1.6160310277957337,
     "total_flos": 5.435725490631475e+20,
-    "train_loss": 0.2802161669254303,
-    "train_runtime": 36290.7707,
-    "train_samples": 121784,
-    "train_samples_per_second": 4.409,
-    "train_steps_per_second": 0.138
+    "train_loss": 0.17819489703178407,
+    "train_runtime": 33619.5535,
+    "train_samples": 99000,
+    "train_samples_per_second": 4.759,
+    "train_steps_per_second": 0.149
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.3137151865475565,
+  "epoch": 1.6160310277957337,
   "eval_steps": 1000,
   "global_step": 5000,
   "is_hyper_param_search": false,
@@ -9,1458 +9,1458 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.006568575932737782,
-      "grad_norm": 4.938570976257324,
+      "epoch": 0.00808015513897867,
+      "grad_norm": 3.75984263420105,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 1.1658,
+      "loss": 0.8538,
       "step": 25
     },
     {
-      "epoch": 0.013137151865475564,
-      "grad_norm": 3.4416959285736084,
+      "epoch": 0.01616031027795734,
+      "grad_norm": 3.785407066345215,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.9316,
+      "loss": 0.7059,
       "step": 50
     },
     {
-      "epoch": 0.01970572779821335,
-      "grad_norm": 3.3691599369049072,
+      "epoch": 0.024240465416936006,
+      "grad_norm": 2.6398749351501465,
       "learning_rate": 1.5e-06,
-      "loss": 0.753,
+      "loss": 0.5499,
       "step": 75
     },
     {
-      "epoch": 0.02627430373095113,
-      "grad_norm": 3.2769742012023926,
+      "epoch": 0.03232062055591468,
+      "grad_norm": 3.0633111000061035,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.6576,
+      "loss": 0.4899,
       "step": 100
     },
     {
-      "epoch": 0.032842879663688915,
-      "grad_norm": 2.5112249851226807,
+      "epoch": 0.040400775694893344,
+      "grad_norm": 2.5752687454223633,
       "learning_rate": 2.5e-06,
-      "loss": 0.6188,
+      "loss": 0.4542,
       "step": 125
     },
     {
-      "epoch": 0.0394114555964267,
-      "grad_norm": 3.0561046600341797,
+      "epoch": 0.04848093083387201,
+      "grad_norm": 2.7746145725250244,
       "learning_rate": 3e-06,
-      "loss": 0.6021,
+      "loss": 0.4448,
       "step": 150
     },
     {
-      "epoch": 0.045980031529164474,
-      "grad_norm": 2.563472032546997,
+      "epoch": 0.05656108597285068,
+      "grad_norm": 2.5188117027282715,
       "learning_rate": 3.5e-06,
-      "loss": 0.564,
+      "loss": 0.4047,
       "step": 175
     },
     {
-      "epoch": 0.05254860746190226,
-      "grad_norm": 2.6831862926483154,
+      "epoch": 0.06464124111182935,
+      "grad_norm": 2.549173593521118,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.5314,
+      "loss": 0.3918,
       "step": 200
     },
     {
-      "epoch": 0.05911718339464004,
-      "grad_norm": 2.383794069290161,
+      "epoch": 0.07272139625080802,
+      "grad_norm": 2.651440382003784,
       "learning_rate": 4.5e-06,
-      "loss": 0.5376,
+      "loss": 0.3738,
       "step": 225
     },
     {
-      "epoch": 0.06568575932737783,
-      "grad_norm": 2.8369593620300293,
+      "epoch": 0.08080155138978669,
+      "grad_norm": 2.6359574794769287,
       "learning_rate": 5e-06,
-      "loss": 0.5404,
+      "loss": 0.3923,
       "step": 250
     },
     {
-      "epoch": 0.07225433526011561,
-      "grad_norm": 2.641814708709717,
+      "epoch": 0.08888170652876536,
+      "grad_norm": 3.015693426132202,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.5014,
+      "loss": 0.3597,
       "step": 275
     },
     {
-      "epoch": 0.0788229111928534,
-      "grad_norm": 2.549522638320923,
+      "epoch": 0.09696186166774402,
+      "grad_norm": 2.4634430408477783,
       "learning_rate": 6e-06,
-      "loss": 0.4978,
+      "loss": 0.3577,
       "step": 300
     },
     {
-      "epoch": 0.08539148712559118,
-      "grad_norm": 2.770048141479492,
+      "epoch": 0.10504201680672269,
+      "grad_norm": 2.6268017292022705,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.5135,
+      "loss": 0.3397,
       "step": 325
     },
     {
-      "epoch": 0.09196006305832895,
-      "grad_norm": 3.170330286026001,
+      "epoch": 0.11312217194570136,
+      "grad_norm": 2.1071865558624268,
       "learning_rate": 7e-06,
-      "loss": 0.504,
+      "loss": 0.344,
       "step": 350
     },
     {
-      "epoch": 0.09852863899106673,
-      "grad_norm": 2.712066650390625,
+      "epoch": 0.12120232708468003,
+      "grad_norm": 2.0719916820526123,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.451,
+      "loss": 0.3232,
       "step": 375
     },
     {
-      "epoch": 0.10509721492380451,
-      "grad_norm": 2.3632166385650635,
+      "epoch": 0.1292824822236587,
+      "grad_norm": 2.445434331893921,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.4533,
+      "loss": 0.3488,
       "step": 400
     },
     {
-      "epoch": 0.1116657908565423,
-      "grad_norm": 2.728026866912842,
+      "epoch": 0.13736263736263737,
+      "grad_norm": 2.1177170276641846,
       "learning_rate": 8.5e-06,
-      "loss": 0.4732,
+      "loss": 0.327,
       "step": 425
     },
     {
-      "epoch": 0.11823436678928008,
-      "grad_norm": 2.7299931049346924,
+      "epoch": 0.14544279250161604,
+      "grad_norm": 2.2949392795562744,
       "learning_rate": 9e-06,
-      "loss": 0.4255,
+      "loss": 0.3228,
       "step": 450
     },
     {
-      "epoch": 0.12480294272201786,
-      "grad_norm": 2.4892563819885254,
+      "epoch": 0.1535229476405947,
+      "grad_norm": 1.701798439025879,
       "learning_rate": 9.5e-06,
-      "loss": 0.4254,
+      "loss": 0.2907,
       "step": 475
     },
     {
-      "epoch": 0.13137151865475566,
-      "grad_norm": 2.3553197383880615,
+      "epoch": 0.16160310277957338,
+      "grad_norm": 1.857657790184021,
       "learning_rate": 1e-05,
-      "loss": 0.4218,
+      "loss": 0.3021,
       "step": 500
     },
     {
-      "epoch": 0.13794009458749343,
-      "grad_norm": 2.1275930404663086,
+      "epoch": 0.16968325791855204,
+      "grad_norm": 1.8236263990402222,
       "learning_rate": 9.944444444444445e-06,
-      "loss": 0.4159,
+      "loss": 0.2878,
       "step": 525
     },
     {
-      "epoch": 0.14450867052023122,
-      "grad_norm": 2.166019916534424,
+      "epoch": 0.1777634130575307,
+      "grad_norm": 2.30806303024292,
       "learning_rate": 9.88888888888889e-06,
-      "loss": 0.4051,
+      "loss": 0.2741,
       "step": 550
     },
     {
-      "epoch": 0.151077246452969,
-      "grad_norm": 2.5494415760040283,
+      "epoch": 0.18584356819650938,
+      "grad_norm": 2.0603034496307373,
       "learning_rate": 9.833333333333333e-06,
-      "loss": 0.399,
+      "loss": 0.2953,
       "step": 575
     },
     {
-      "epoch": 0.1576458223857068,
-      "grad_norm": 2.482174873352051,
+      "epoch": 0.19392372333548805,
+      "grad_norm": 2.1043479442596436,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 0.4013,
+      "loss": 0.2855,
       "step": 600
     },
     {
-      "epoch": 0.16421439831844456,
-      "grad_norm": 2.3230655193328857,
+      "epoch": 0.2020038784744667,
+      "grad_norm": 1.921908974647522,
       "learning_rate": 9.722222222222223e-06,
-      "loss": 0.3736,
+      "loss": 0.2709,
       "step": 625
     },
     {
-      "epoch": 0.17078297425118236,
-      "grad_norm": 2.1272637844085693,
+      "epoch": 0.21008403361344538,
+      "grad_norm": 2.1535732746124268,
       "learning_rate": 9.666666666666667e-06,
-      "loss": 0.3715,
+      "loss": 0.251,
       "step": 650
     },
     {
-      "epoch": 0.17735155018392013,
-      "grad_norm": 2.2144370079040527,
+      "epoch": 0.21816418875242405,
+      "grad_norm": 2.01731014251709,
       "learning_rate": 9.611111111111112e-06,
-      "loss": 0.389,
+      "loss": 0.2817,
       "step": 675
     },
     {
-      "epoch": 0.1839201261166579,
-      "grad_norm": 2.156562566757202,
+      "epoch": 0.22624434389140272,
+      "grad_norm": 1.9176462888717651,
       "learning_rate": 9.555555555555556e-06,
-      "loss": 0.3779,
+      "loss": 0.2608,
       "step": 700
     },
     {
-      "epoch": 0.1904887020493957,
-      "grad_norm": 1.9418479204177856,
+      "epoch": 0.23432449903038138,
+      "grad_norm": 2.2685463428497314,
       "learning_rate": 9.5e-06,
-      "loss": 0.3624,
+      "loss": 0.2584,
       "step": 725
     },
     {
-      "epoch": 0.19705727798213346,
-      "grad_norm": 1.9420461654663086,
+      "epoch": 0.24240465416936005,
+      "grad_norm": 1.6904027462005615,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 0.3726,
+      "loss": 0.2529,
       "step": 750
     },
     {
-      "epoch": 0.20362585391487126,
-      "grad_norm": 1.929376482963562,
+      "epoch": 0.2504848093083387,
+      "grad_norm": 2.014220714569092,
       "learning_rate": 9.38888888888889e-06,
-      "loss": 0.3501,
+      "loss": 0.2443,
       "step": 775
     },
     {
-      "epoch": 0.21019442984760903,
-      "grad_norm": 2.257051467895508,
+      "epoch": 0.2585649644473174,
+      "grad_norm": 2.1156768798828125,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 0.3755,
+      "loss": 0.2548,
       "step": 800
     },
     {
-      "epoch": 0.21676300578034682,
-      "grad_norm": 2.102417230606079,
+      "epoch": 0.26664511958629605,
+      "grad_norm": 2.0581390857696533,
       "learning_rate": 9.277777777777778e-06,
-      "loss": 0.3671,
+      "loss": 0.2391,
       "step": 825
     },
     {
-      "epoch": 0.2233315817130846,
-      "grad_norm": 2.096370220184326,
+      "epoch": 0.27472527472527475,
+      "grad_norm": 1.9932626485824585,
       "learning_rate": 9.222222222222224e-06,
-      "loss": 0.3397,
+      "loss": 0.2326,
       "step": 850
     },
     {
-      "epoch": 0.2299001576458224,
-      "grad_norm": 2.0005109310150146,
+      "epoch": 0.2828054298642534,
+      "grad_norm": 1.7348275184631348,
       "learning_rate": 9.166666666666666e-06,
-      "loss": 0.3467,
+      "loss": 0.2399,
       "step": 875
     },
     {
-      "epoch": 0.23646873357856016,
-      "grad_norm": 2.1975176334381104,
+      "epoch": 0.2908855850032321,
+      "grad_norm": 1.7778377532958984,
       "learning_rate": 9.111111111111112e-06,
-      "loss": 0.3341,
+      "loss": 0.2272,
       "step": 900
     },
     {
-      "epoch": 0.24303730951129796,
-      "grad_norm": 2.0950164794921875,
+      "epoch": 0.2989657401422107,
+      "grad_norm": 1.917076826095581,
       "learning_rate": 9.055555555555556e-06,
-      "loss": 0.3398,
+      "loss": 0.2498,
       "step": 925
     },
     {
-      "epoch": 0.24960588544403572,
-      "grad_norm": 2.0137791633605957,
+      "epoch": 0.3070458952811894,
+      "grad_norm": 1.727513313293457,
       "learning_rate": 9e-06,
-      "loss": 0.3183,
+      "loss": 0.226,
       "step": 950
     },
     {
-      "epoch": 0.2561744613767735,
-      "grad_norm": 1.97981595993042,
+      "epoch": 0.31512605042016806,
+      "grad_norm": 2.011462688446045,
       "learning_rate": 8.944444444444446e-06,
-      "loss": 0.3653,
+      "loss": 0.2358,
       "step": 975
     },
     {
-      "epoch": 0.2627430373095113,
-      "grad_norm": 1.9213696718215942,
+      "epoch": 0.32320620555914675,
+      "grad_norm": 2.0421934127807617,
       "learning_rate": 8.888888888888888e-06,
-      "loss": 0.3429,
+      "loss": 0.2469,
       "step": 1000
     },
     {
-      "epoch": 0.2627430373095113,
-      "eval_loss": 0.3295910656452179,
-      "eval_runtime": 594.427,
-      "eval_samples_per_second": 2.052,
-      "eval_steps_per_second": 0.13,
-      "eval_wer": 0.2299665865525529,
+      "epoch": 0.32320620555914675,
+      "eval_loss": 0.23132415115833282,
+      "eval_runtime": 421.4559,
+      "eval_samples_per_second": 2.373,
+      "eval_steps_per_second": 0.149,
+      "eval_wer": 0.17325433135828397,
       "step": 1000
     },
     {
-      "epoch": 0.26931161324224906,
-      "grad_norm": 2.122868299484253,
+      "epoch": 0.3312863606981254,
+      "grad_norm": 2.1990480422973633,
       "learning_rate": 8.833333333333334e-06,
-      "loss": 0.318,
+      "loss": 0.2397,
       "step": 1025
     },
     {
-      "epoch": 0.27588018917498686,
-      "grad_norm": 2.259373426437378,
+      "epoch": 0.3393665158371041,
+      "grad_norm": 1.6808555126190186,
       "learning_rate": 8.777777777777778e-06,
-      "loss": 0.3364,
+      "loss": 0.212,
       "step": 1050
     },
     {
-      "epoch": 0.28244876510772465,
-      "grad_norm": 1.978367567062378,
+      "epoch": 0.3474466709760827,
+      "grad_norm": 1.9749352931976318,
       "learning_rate": 8.722222222222224e-06,
-      "loss": 0.3125,
+      "loss": 0.2148,
       "step": 1075
     },
     {
-      "epoch": 0.28901734104046245,
-      "grad_norm": 1.8453904390335083,
+      "epoch": 0.3555268261150614,
+      "grad_norm": 2.0048928260803223,
       "learning_rate": 8.666666666666668e-06,
-      "loss": 0.3009,
+      "loss": 0.2161,
       "step": 1100
     },
     {
-      "epoch": 0.2955859169732002,
-      "grad_norm": 2.002053737640381,
+      "epoch": 0.36360698125404006,
+      "grad_norm": 1.8271081447601318,
       "learning_rate": 8.611111111111112e-06,
-      "loss": 0.3087,
+      "loss": 0.237,
       "step": 1125
     },
     {
-      "epoch": 0.302154492905938,
-      "grad_norm": 2.080690860748291,
+      "epoch": 0.37168713639301876,
+      "grad_norm": 1.751715898513794,
       "learning_rate": 8.555555555555556e-06,
-      "loss": 0.3188,
+      "loss": 0.2088,
       "step": 1150
     },
     {
-      "epoch": 0.3087230688386758,
-      "grad_norm": 1.6879727840423584,
+      "epoch": 0.3797672915319974,
+      "grad_norm": 1.490201473236084,
       "learning_rate": 8.5e-06,
-      "loss": 0.3043,
+      "loss": 0.2207,
       "step": 1175
     },
     {
-      "epoch": 0.3152916447714136,
-      "grad_norm": 2.2736353874206543,
+      "epoch": 0.3878474466709761,
+      "grad_norm": 2.108602285385132,
       "learning_rate": 8.444444444444446e-06,
-      "loss": 0.3165,
+      "loss": 0.2177,
       "step": 1200
     },
     {
-      "epoch": 0.3218602207041513,
-      "grad_norm": 1.8325968980789185,
+      "epoch": 0.39592760180995473,
+      "grad_norm": 1.637518048286438,
       "learning_rate": 8.38888888888889e-06,
-      "loss": 0.3236,
+      "loss": 0.2134,
       "step": 1225
     },
     {
-      "epoch": 0.3284287966368891,
-      "grad_norm": 1.8235770463943481,
+      "epoch": 0.4040077569489334,
+      "grad_norm": 1.8813297748565674,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.3057,
+      "loss": 0.2048,
       "step": 1250
     },
     {
-      "epoch": 0.3349973725696269,
-      "grad_norm": 2.09566068649292,
+      "epoch": 0.41208791208791207,
+      "grad_norm": 1.61934494972229,
       "learning_rate": 8.277777777777778e-06,
-      "loss": 0.2845,
+      "loss": 0.1951,
       "step": 1275
     },
     {
-      "epoch": 0.3415659485023647,
-      "grad_norm": 1.9097365140914917,
+      "epoch": 0.42016806722689076,
+      "grad_norm": 1.9593381881713867,
       "learning_rate": 8.222222222222222e-06,
-      "loss": 0.2833,
+      "loss": 0.1989,
       "step": 1300
     },
     {
-      "epoch": 0.34813452443510245,
-      "grad_norm": 1.9709367752075195,
+      "epoch": 0.4282482223658694,
+      "grad_norm": 1.7399969100952148,
       "learning_rate": 8.166666666666668e-06,
-      "loss": 0.294,
+      "loss": 0.1837,
       "step": 1325
     },
     {
-      "epoch": 0.35470310036784025,
-      "grad_norm": 1.9940749406814575,
+      "epoch": 0.4363283775048481,
+      "grad_norm": 1.9102469682693481,
       "learning_rate": 8.111111111111112e-06,
-      "loss": 0.2789,
+      "loss": 0.19,
       "step": 1350
     },
     {
-      "epoch": 0.36127167630057805,
-      "grad_norm": 2.122657299041748,
+      "epoch": 0.44440853264382674,
+      "grad_norm": 1.8125574588775635,
       "learning_rate": 8.055555555555557e-06,
-      "loss": 0.3026,
+      "loss": 0.2015,
       "step": 1375
     },
     {
-      "epoch": 0.3678402522333158,
-      "grad_norm": 1.742019772529602,
+      "epoch": 0.45248868778280543,
+      "grad_norm": 1.8675082921981812,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.2843,
+      "loss": 0.2076,
       "step": 1400
     },
     {
-      "epoch": 0.3744088281660536,
-      "grad_norm": 1.8771026134490967,
+      "epoch": 0.46056884292178407,
+      "grad_norm": 2.056074619293213,
       "learning_rate": 7.944444444444445e-06,
-      "loss": 0.2937,
+      "loss": 0.2054,
       "step": 1425
     },
     {
-      "epoch": 0.3809774040987914,
-      "grad_norm": 1.617311954498291,
+      "epoch": 0.46864899806076277,
+      "grad_norm": 1.4601351022720337,
       "learning_rate": 7.88888888888889e-06,
-      "loss": 0.2749,
+      "loss": 0.2057,
       "step": 1450
     },
     {
-      "epoch": 0.3875459800315292,
-      "grad_norm": 1.902928113937378,
+      "epoch": 0.47672915319974146,
+      "grad_norm": 1.6205873489379883,
       "learning_rate": 7.833333333333333e-06,
-      "loss": 0.2929,
+      "loss": 0.1887,
       "step": 1475
     },
     {
-      "epoch": 0.3941145559642669,
-      "grad_norm": 1.9772895574569702,
+      "epoch": 0.4848093083387201,
+      "grad_norm": 1.356102705001831,
       "learning_rate": 7.77777777777778e-06,
-      "loss": 0.2699,
+      "loss": 0.1767,
       "step": 1500
     },
     {
-      "epoch": 0.4006831318970047,
-      "grad_norm": 1.7294281721115112,
+      "epoch": 0.4928894634776988,
+      "grad_norm": 2.1016225814819336,
       "learning_rate": 7.722222222222223e-06,
-      "loss": 0.288,
+      "loss": 0.1859,
       "step": 1525
     },
     {
-      "epoch": 0.4072517078297425,
-      "grad_norm": 1.6163759231567383,
+      "epoch": 0.5009696186166774,
+      "grad_norm": 1.3985319137573242,
       "learning_rate": 7.666666666666667e-06,
-      "loss": 0.2919,
+      "loss": 0.1981,
       "step": 1550
     },
     {
-      "epoch": 0.4138202837624803,
-      "grad_norm": 1.9823168516159058,
+      "epoch": 0.5090497737556561,
+      "grad_norm": 1.8482751846313477,
       "learning_rate": 7.611111111111111e-06,
-      "loss": 0.2691,
+      "loss": 0.1979,
       "step": 1575
     },
     {
-      "epoch": 0.42038885969521805,
-      "grad_norm": 2.085510730743408,
+      "epoch": 0.5171299288946348,
+      "grad_norm": 1.6422449350357056,
       "learning_rate": 7.555555555555556e-06,
-      "loss": 0.2807,
+      "loss": 0.1858,
       "step": 1600
     },
     {
-      "epoch": 0.42695743562795585,
-      "grad_norm": 2.1536219120025635,
+      "epoch": 0.5252100840336135,
+      "grad_norm": 1.5610370635986328,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.2977,
+      "loss": 0.1988,
       "step": 1625
     },
     {
-      "epoch": 0.43352601156069365,
-      "grad_norm": 2.23018741607666,
+      "epoch": 0.5332902391725921,
+      "grad_norm": 1.5282772779464722,
       "learning_rate": 7.444444444444445e-06,
-      "loss": 0.291,
+      "loss": 0.1904,
       "step": 1650
     },
     {
-      "epoch": 0.44009458749343144,
-      "grad_norm": 1.8097656965255737,
+      "epoch": 0.5413703943115707,
+      "grad_norm": 1.5836628675460815,
       "learning_rate": 7.38888888888889e-06,
-      "loss": 0.3125,
+      "loss": 0.1913,
       "step": 1675
     },
     {
-      "epoch": 0.4466631634261692,
-      "grad_norm": 1.557788372039795,
+      "epoch": 0.5494505494505495,
+      "grad_norm": 1.7465559244155884,
       "learning_rate": 7.333333333333333e-06,
-      "loss": 0.2743,
+      "loss": 0.1638,
       "step": 1700
     },
     {
-      "epoch": 0.453231739358907,
-      "grad_norm": 1.677343726158142,
+      "epoch": 0.5575307045895281,
+      "grad_norm": 1.7590513229370117,
       "learning_rate": 7.277777777777778e-06,
-      "loss": 0.2568,
+      "loss": 0.1946,
       "step": 1725
     },
     {
-      "epoch": 0.4598003152916448,
-      "grad_norm": 1.7231301069259644,
+      "epoch": 0.5656108597285068,
+      "grad_norm": 1.7977447509765625,
       "learning_rate": 7.222222222222223e-06,
-      "loss": 0.2701,
+      "loss": 0.1919,
       "step": 1750
     },
     {
-      "epoch": 0.4663688912243826,
-      "grad_norm": 1.632379412651062,
+      "epoch": 0.5736910148674854,
+      "grad_norm": 1.5070719718933105,
       "learning_rate": 7.166666666666667e-06,
-      "loss": 0.2795,
+      "loss": 0.1801,
       "step": 1775
     },
     {
-      "epoch": 0.4729374671571203,
-      "grad_norm": 2.042989730834961,
+      "epoch": 0.5817711700064642,
+      "grad_norm": 1.7022117376327515,
       "learning_rate": 7.111111111111112e-06,
-      "loss": 0.2714,
+      "loss": 0.1831,
       "step": 1800
     },
     {
-      "epoch": 0.4795060430898581,
-      "grad_norm": 1.8419965505599976,
+      "epoch": 0.5898513251454428,
+      "grad_norm": 1.6493513584136963,
       "learning_rate": 7.055555555555557e-06,
-      "loss": 0.2638,
+      "loss": 0.1889,
       "step": 1825
     },
     {
-      "epoch": 0.4860746190225959,
-      "grad_norm": 1.7692322731018066,
+      "epoch": 0.5979314802844214,
+      "grad_norm": 1.4757349491119385,
       "learning_rate": 7e-06,
-      "loss": 0.2631,
+      "loss": 0.1834,
       "step": 1850
     },
     {
-      "epoch": 0.4926431949553337,
-      "grad_norm": 1.8727885484695435,
+      "epoch": 0.6060116354234001,
+      "grad_norm": 1.7376744747161865,
       "learning_rate": 6.944444444444445e-06,
-      "loss": 0.2647,
+      "loss": 0.1676,
       "step": 1875
     },
     {
-      "epoch": 0.49921177088807145,
-      "grad_norm": 1.5731993913650513,
+      "epoch": 0.6140917905623788,
+      "grad_norm": 1.9608112573623657,
       "learning_rate": 6.88888888888889e-06,
-      "loss": 0.2765,
+      "loss": 0.1884,
       "step": 1900
     },
     {
-      "epoch": 0.5057803468208093,
-      "grad_norm": 1.9374446868896484,
+      "epoch": 0.6221719457013575,
+      "grad_norm": 1.412118673324585,
       "learning_rate": 6.833333333333334e-06,
-      "loss": 0.2672,
+      "loss": 0.1949,
       "step": 1925
     },
     {
-      "epoch": 0.512348922753547,
-      "grad_norm": 1.702355980873108,
+      "epoch": 0.6302521008403361,
+      "grad_norm": 2.119964361190796,
       "learning_rate": 6.777777777777779e-06,
-      "loss": 0.2574,
+      "loss": 0.193,
       "step": 1950
     },
     {
-      "epoch": 0.5189174986862848,
-      "grad_norm": 1.539027214050293,
+      "epoch": 0.6383322559793148,
+      "grad_norm": 1.9113435745239258,
       "learning_rate": 6.7222222222222235e-06,
-      "loss": 0.2651,
+      "loss": 0.1662,
       "step": 1975
     },
     {
-      "epoch": 0.5254860746190226,
-      "grad_norm": 1.8967093229293823,
+      "epoch": 0.6464124111182935,
+      "grad_norm": 1.3223708868026733,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.2636,
+      "loss": 0.1578,
       "step": 2000
     },
     {
-      "epoch": 0.5254860746190226,
-      "eval_loss": 0.2615206837654114,
-      "eval_runtime": 532.8353,
-      "eval_samples_per_second": 2.29,
-      "eval_steps_per_second": 0.145,
-      "eval_wer": 0.1925669734451023,
+      "epoch": 0.6464124111182935,
+      "eval_loss": 0.1784905195236206,
+      "eval_runtime": 427.4993,
+      "eval_samples_per_second": 2.339,
+      "eval_steps_per_second": 0.147,
+      "eval_wer": 0.13732843321083027,
       "step": 2000
     },
     {
-      "epoch": 0.5320546505517604,
-      "grad_norm": 1.688244342803955,
+      "epoch": 0.6544925662572721,
+      "grad_norm": 1.5342998504638672,
       "learning_rate": 6.6111111111111115e-06,
-      "loss": 0.2683,
+      "loss": 0.1727,
       "step": 2025
     },
     {
-      "epoch": 0.5386232264844981,
-      "grad_norm": 1.8095322847366333,
+      "epoch": 0.6625727213962508,
+      "grad_norm": 1.322567105293274,
       "learning_rate": 6.555555555555556e-06,
-      "loss": 0.2707,
+      "loss": 0.1858,
       "step": 2050
     },
     {
-      "epoch": 0.545191802417236,
-      "grad_norm": 1.8945379257202148,
+      "epoch": 0.6706528765352294,
+      "grad_norm": 1.5235811471939087,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.2535,
+      "loss": 0.1523,
       "step": 2075
     },
     {
-      "epoch": 0.5517603783499737,
-      "grad_norm": 1.9910788536071777,
+      "epoch": 0.6787330316742082,
+      "grad_norm": 1.478602647781372,
       "learning_rate": 6.444444444444445e-06,
-      "loss": 0.2486,
+      "loss": 0.1667,
       "step": 2100
     },
     {
-      "epoch": 0.5583289542827115,
-      "grad_norm": 1.3568848371505737,
+      "epoch": 0.6868131868131868,
+      "grad_norm": 1.6120604276657104,
       "learning_rate": 6.3888888888888885e-06,
-      "loss": 0.2755,
+      "loss": 0.163,
       "step": 2125
     },
     {
-      "epoch": 0.5648975302154493,
-      "grad_norm": 1.7561497688293457,
+      "epoch": 0.6948933419521655,
+      "grad_norm": 1.3464877605438232,
       "learning_rate": 6.333333333333333e-06,
-      "loss": 0.2679,
+      "loss": 0.1624,
       "step": 2150
     },
     {
-      "epoch": 0.571466106148187,
-      "grad_norm": 1.6052385568618774,
+      "epoch": 0.7029734970911441,
+      "grad_norm": 1.2172194719314575,
       "learning_rate": 6.277777777777778e-06,
-      "loss": 0.2497,
+      "loss": 0.1483,
       "step": 2175
     },
     {
-      "epoch": 0.5780346820809249,
-      "grad_norm": 1.5882091522216797,
+      "epoch": 0.7110536522301228,
+      "grad_norm": 1.5536859035491943,
       "learning_rate": 6.222222222222223e-06,
-      "loss": 0.2673,
+      "loss": 0.1552,
       "step": 2200
     },
     {
-      "epoch": 0.5846032580136626,
-      "grad_norm": 1.7540193796157837,
+      "epoch": 0.7191338073691015,
+      "grad_norm": 1.394464373588562,
       "learning_rate": 6.166666666666667e-06,
-      "loss": 0.2335,
+      "loss": 0.1536,
       "step": 2225
     },
     {
-      "epoch": 0.5911718339464004,
-      "grad_norm": 1.4882384538650513,
+      "epoch": 0.7272139625080801,
+      "grad_norm": 1.8853507041931152,
       "learning_rate": 6.111111111111112e-06,
-      "loss": 0.2555,
+      "loss": 0.1673,
       "step": 2250
     },
     {
-      "epoch": 0.5977404098791382,
-      "grad_norm": 2.00600266456604,
+      "epoch": 0.7352941176470589,
+      "grad_norm": 1.529552936553955,
       "learning_rate": 6.055555555555555e-06,
-      "loss": 0.2687,
+      "loss": 0.1622,
       "step": 2275
     },
     {
-      "epoch": 0.604308985811876,
-      "grad_norm": 1.4347535371780396,
+      "epoch": 0.7433742727860375,
+      "grad_norm": 1.541446566581726,
       "learning_rate": 6e-06,
-      "loss": 0.2681,
+      "loss": 0.1647,
       "step": 2300
     },
     {
-      "epoch": 0.6108775617446137,
-      "grad_norm": 1.8053362369537354,
+      "epoch": 0.7514544279250162,
+      "grad_norm": 1.6299506425857544,
       "learning_rate": 5.944444444444445e-06,
-      "loss": 0.2568,
+      "loss": 0.172,
       "step": 2325
     },
     {
-      "epoch": 0.6174461376773516,
-      "grad_norm": 1.657359004020691,
+      "epoch": 0.7595345830639948,
+      "grad_norm": 1.5410858392715454,
       "learning_rate": 5.88888888888889e-06,
-      "loss": 0.246,
+      "loss": 0.1558,
       "step": 2350
     },
     {
-      "epoch": 0.6240147136100893,
-      "grad_norm": 1.6917238235473633,
+      "epoch": 0.7676147382029735,
+      "grad_norm": 1.5796810388565063,
       "learning_rate": 5.833333333333334e-06,
-      "loss": 0.2333,
+      "loss": 0.158,
       "step": 2375
     },
     {
-      "epoch": 0.6305832895428272,
-      "grad_norm": 2.0534191131591797,
+      "epoch": 0.7756948933419522,
+      "grad_norm": 1.5637125968933105,
       "learning_rate": 5.777777777777778e-06,
-      "loss": 0.2557,
+      "loss": 0.1583,
       "step": 2400
     },
     {
-      "epoch": 0.6371518654755649,
-      "grad_norm": 1.6997853517532349,
+      "epoch": 0.7837750484809308,
+      "grad_norm": 1.4852577447891235,
       "learning_rate": 5.722222222222222e-06,
-      "loss": 0.2888,
+      "loss": 0.1618,
       "step": 2425
     },
     {
-      "epoch": 0.6437204414083026,
-      "grad_norm": 1.7183536291122437,
+      "epoch": 0.7918552036199095,
+      "grad_norm": 1.6491366624832153,
       "learning_rate": 5.666666666666667e-06,
-      "loss": 0.2364,
+      "loss": 0.16,
       "step": 2450
     },
     {
-      "epoch": 0.6502890173410405,
-      "grad_norm": 1.6452535390853882,
+      "epoch": 0.7999353587588882,
+      "grad_norm": 1.7685949802398682,
       "learning_rate": 5.611111111111112e-06,
-      "loss": 0.2395,
+      "loss": 0.1706,
       "step": 2475
     },
     {
-      "epoch": 0.6568575932737782,
-      "grad_norm": 1.7632161378860474,
+      "epoch": 0.8080155138978669,
+      "grad_norm": 1.4126862287521362,
       "learning_rate": 5.555555555555557e-06,
-      "loss": 0.2547,
+      "loss": 0.1525,
       "step": 2500
     },
     {
-      "epoch": 0.663426169206516,
-      "grad_norm": 1.9056440591812134,
+      "epoch": 0.8160956690368455,
+      "grad_norm": 1.4942584037780762,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.249,
+      "loss": 0.1628,
       "step": 2525
     },
     {
-      "epoch": 0.6699947451392538,
-      "grad_norm": 2.083576202392578,
+      "epoch": 0.8241758241758241,
+      "grad_norm": 1.6680798530578613,
       "learning_rate": 5.444444444444445e-06,
-      "loss": 0.2596,
+      "loss": 0.1647,
       "step": 2550
     },
     {
-      "epoch": 0.6765633210719916,
-      "grad_norm": 2.1211483478546143,
+      "epoch": 0.8322559793148029,
+      "grad_norm": 1.1703355312347412,
       "learning_rate": 5.388888888888889e-06,
-      "loss": 0.2435,
+      "loss": 0.1519,
       "step": 2575
     },
     {
-      "epoch": 0.6831318970047294,
-      "grad_norm": 1.599048376083374,
+      "epoch": 0.8403361344537815,
+      "grad_norm": 1.4046279191970825,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 0.2376,
+      "loss": 0.1492,
       "step": 2600
     },
     {
-      "epoch": 0.6897004729374672,
-      "grad_norm": 1.6461100578308105,
+      "epoch": 0.8484162895927602,
+      "grad_norm": 1.4554412364959717,
       "learning_rate": 5.2777777777777785e-06,
-      "loss": 0.2457,
+      "loss": 0.1452,
       "step": 2625
     },
     {
-      "epoch": 0.6962690488702049,
-      "grad_norm": 1.660069227218628,
+      "epoch": 0.8564964447317388,
+      "grad_norm": 1.737576961517334,
       "learning_rate": 5.2222222222222226e-06,
-      "loss": 0.2354,
+      "loss": 0.1629,
       "step": 2650
     },
     {
-      "epoch": 0.7028376248029428,
-      "grad_norm": 2.221201181411743,
+      "epoch": 0.8645765998707176,
+      "grad_norm": 1.261842966079712,
       "learning_rate": 5.1666666666666675e-06,
-      "loss": 0.2794,
+      "loss": 0.1539,
       "step": 2675
     },
     {
-      "epoch": 0.7094062007356805,
-      "grad_norm": 2.4395995140075684,
+      "epoch": 0.8726567550096962,
+      "grad_norm": 1.5773130655288696,
       "learning_rate": 5.1111111111111115e-06,
-      "loss": 0.2499,
+      "loss": 0.1434,
       "step": 2700
     },
     {
-      "epoch": 0.7159747766684182,
-      "grad_norm": 1.4763926267623901,
+      "epoch": 0.8807369101486748,
+      "grad_norm": 1.619884967803955,
       "learning_rate": 5.0555555555555555e-06,
-      "loss": 0.2783,
+      "loss": 0.1753,
       "step": 2725
     },
     {
-      "epoch": 0.7225433526011561,
-      "grad_norm": 1.6573847532272339,
+      "epoch": 0.8888170652876535,
+      "grad_norm": 1.175441861152649,
       "learning_rate": 5e-06,
-      "loss": 0.2349,
+      "loss": 0.1594,
       "step": 2750
     },
     {
-      "epoch": 0.7291119285338938,
-      "grad_norm": 1.8241550922393799,
+      "epoch": 0.8968972204266322,
+      "grad_norm": 1.8705310821533203,
       "learning_rate": 4.944444444444445e-06,
-      "loss": 0.2449,
+      "loss": 0.1618,
       "step": 2775
     },
     {
-      "epoch": 0.7356805044666316,
-      "grad_norm": 1.6332080364227295,
+      "epoch": 0.9049773755656109,
+      "grad_norm": 1.5610700845718384,
       "learning_rate": 4.888888888888889e-06,
-      "loss": 0.2413,
+      "loss": 0.1537,
       "step": 2800
     },
     {
-      "epoch": 0.7422490803993694,
-      "grad_norm": 1.626541256904602,
+      "epoch": 0.9130575307045895,
+      "grad_norm": 1.940385103225708,
       "learning_rate": 4.833333333333333e-06,
-      "loss": 0.2499,
+      "loss": 0.1602,
       "step": 2825
     },
     {
-      "epoch": 0.7488176563321072,
-      "grad_norm": 1.5557328462600708,
+      "epoch": 0.9211376858435681,
+      "grad_norm": 1.4349099397659302,
       "learning_rate": 4.777777777777778e-06,
-      "loss": 0.2534,
+      "loss": 0.1492,
       "step": 2850
     },
     {
-      "epoch": 0.755386232264845,
-      "grad_norm": 1.932283878326416,
+      "epoch": 0.9292178409825469,
+      "grad_norm": 1.7276194095611572,
       "learning_rate": 4.722222222222222e-06,
-      "loss": 0.2379,
+      "loss": 0.1418,
       "step": 2875
     },
     {
-      "epoch": 0.7619548081975828,
-      "grad_norm": 1.9064332246780396,
+      "epoch": 0.9372979961215255,
+      "grad_norm": 1.138739824295044,
       "learning_rate": 4.666666666666667e-06,
-      "loss": 0.2323,
+      "loss": 0.1566,
       "step": 2900
     },
     {
-      "epoch": 0.7685233841303205,
-      "grad_norm": 1.5515189170837402,
+      "epoch": 0.9453781512605042,
+      "grad_norm": 1.1318280696868896,
       "learning_rate": 4.611111111111112e-06,
-      "loss": 0.2328,
+      "loss": 0.1489,
       "step": 2925
     },
     {
-      "epoch": 0.7750919600630584,
-      "grad_norm": 2.1047863960266113,
+      "epoch": 0.9534583063994829,
+      "grad_norm": 1.766103744506836,
       "learning_rate": 4.555555555555556e-06,
-      "loss": 0.248,
+      "loss": 0.1567,
       "step": 2950
     },
     {
-      "epoch": 0.7816605359957961,
-      "grad_norm": 1.8366841077804565,
+      "epoch": 0.9615384615384616,
+      "grad_norm": 1.9743990898132324,
       "learning_rate": 4.5e-06,
-      "loss": 0.2584,
+      "loss": 0.1524,
       "step": 2975
     },
     {
-      "epoch": 0.7882291119285338,
-      "grad_norm": 1.949874997138977,
+      "epoch": 0.9696186166774402,
+      "grad_norm": 1.7132006883621216,
       "learning_rate": 4.444444444444444e-06,
-      "loss": 0.2316,
+      "loss": 0.1491,
       "step": 3000
     },
     {
-      "epoch": 0.7882291119285338,
-      "eval_loss": 0.23398828506469727,
-      "eval_runtime": 533.7623,
-      "eval_samples_per_second": 2.286,
-      "eval_steps_per_second": 0.144,
-      "eval_wer": 0.1797877952986693,
+      "epoch": 0.9696186166774402,
+      "eval_loss": 0.15305228531360626,
+      "eval_runtime": 421.6838,
+      "eval_samples_per_second": 2.371,
+      "eval_steps_per_second": 0.149,
+      "eval_wer": 0.12127803195079877,
       "step": 3000
     },
     {
-      "epoch": 0.7947976878612717,
-      "grad_norm": 2.0107836723327637,
+      "epoch": 0.9776987718164188,
+      "grad_norm": 1.675523042678833,
       "learning_rate": 4.388888888888889e-06,
-      "loss": 0.2421,
+      "loss": 0.1359,
       "step": 3025
     },
     {
-      "epoch": 0.8013662637940094,
-      "grad_norm": 1.959560751914978,
+      "epoch": 0.9857789269553976,
+      "grad_norm": 1.1501892805099487,
       "learning_rate": 4.333333333333334e-06,
-      "loss": 0.2319,
+      "loss": 0.1527,
       "step": 3050
     },
     {
-      "epoch": 0.8079348397267473,
-      "grad_norm": 1.6217644214630127,
+      "epoch": 0.9938590820943762,
+      "grad_norm": 1.8712406158447266,
       "learning_rate": 4.277777777777778e-06,
-      "loss": 0.2383,
+      "loss": 0.1572,
       "step": 3075
     },
     {
-      "epoch": 0.814503415659485,
-      "grad_norm": 2.180353879928589,
+      "epoch": 1.0019392372333549,
+      "grad_norm": 1.688395619392395,
       "learning_rate": 4.222222222222223e-06,
-      "loss": 0.2374,
+      "loss": 0.143,
       "step": 3100
     },
     {
-      "epoch": 0.8210719915922228,
-      "grad_norm": 1.7979501485824585,
+      "epoch": 1.0100193923723335,
+      "grad_norm": 1.719205379486084,
       "learning_rate": 4.166666666666667e-06,
-      "loss": 0.2437,
+      "loss": 0.1135,
       "step": 3125
     },
     {
-      "epoch": 0.8276405675249606,
-      "grad_norm": 1.6101832389831543,
+      "epoch": 1.0180995475113122,
+      "grad_norm": 1.399117112159729,
       "learning_rate": 4.111111111111111e-06,
-      "loss": 0.2142,
+      "loss": 0.1138,
       "step": 3150
     },
     {
-      "epoch": 0.8342091434576984,
-      "grad_norm": 2.010740280151367,
+      "epoch": 1.0261797026502908,
+      "grad_norm": 1.3488880395889282,
       "learning_rate": 4.055555555555556e-06,
-      "loss": 0.2489,
+      "loss": 0.097,
       "step": 3175
     },
     {
-      "epoch": 0.8407777193904361,
-      "grad_norm": 1.62699556350708,
+      "epoch": 1.0342598577892697,
+      "grad_norm": 1.3759610652923584,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.2362,
+      "loss": 0.1073,
       "step": 3200
     },
     {
-      "epoch": 0.847346295323174,
-      "grad_norm": 1.6320332288742065,
+      "epoch": 1.0423400129282483,
+      "grad_norm": 1.2929582595825195,
       "learning_rate": 3.944444444444445e-06,
-      "loss": 0.2134,
+      "loss": 0.1067,
       "step": 3225
     },
     {
-      "epoch": 0.8539148712559117,
-      "grad_norm": 1.5167447328567505,
+      "epoch": 1.050420168067227,
+      "grad_norm": 1.2151849269866943,
       "learning_rate": 3.88888888888889e-06,
-      "loss": 0.219,
+      "loss": 0.1023,
       "step": 3250
     },
     {
-      "epoch": 0.8604834471886496,
-      "grad_norm": 1.8277373313903809,
+      "epoch": 1.0585003232062056,
+      "grad_norm": 1.4267935752868652,
       "learning_rate": 3.833333333333334e-06,
-      "loss": 0.2593,
+      "loss": 0.1098,
       "step": 3275
     },
     {
-      "epoch": 0.8670520231213873,
-      "grad_norm": 1.6318010091781616,
+      "epoch": 1.0665804783451842,
+      "grad_norm": 1.4334278106689453,
       "learning_rate": 3.777777777777778e-06,
-      "loss": 0.2151,
+      "loss": 0.1001,
       "step": 3300
     },
     {
-      "epoch": 0.873620599054125,
-      "grad_norm": 1.9252204895019531,
+      "epoch": 1.0746606334841629,
+      "grad_norm": 1.3168810606002808,
       "learning_rate": 3.7222222222222225e-06,
-      "loss": 0.2509,
+      "loss": 0.1025,
       "step": 3325
     },
     {
-      "epoch": 0.8801891749868629,
-      "grad_norm": 2.156442880630493,
+      "epoch": 1.0827407886231415,
+      "grad_norm": 1.4960328340530396,
       "learning_rate": 3.6666666666666666e-06,
-      "loss": 0.2407,
+      "loss": 0.109,
       "step": 3350
     },
     {
-      "epoch": 0.8867577509196006,
-      "grad_norm": 1.9615235328674316,
+      "epoch": 1.0908209437621201,
+      "grad_norm": 1.430653691291809,
       "learning_rate": 3.6111111111111115e-06,
-      "loss": 0.2299,
+      "loss": 0.0938,
       "step": 3375
     },
     {
-      "epoch": 0.8933263268523384,
-      "grad_norm": 1.985635757446289,
+      "epoch": 1.098901098901099,
+      "grad_norm": 1.261281132698059,
       "learning_rate": 3.555555555555556e-06,
-      "loss": 0.2451,
+      "loss": 0.0978,
       "step": 3400
     },
     {
-      "epoch": 0.8998949027850762,
-      "grad_norm": 1.5059348344802856,
+      "epoch": 1.1069812540400776,
+      "grad_norm": 1.2067958116531372,
       "learning_rate": 3.5e-06,
-      "loss": 0.2389,
+      "loss": 0.1008,
       "step": 3425
     },
     {
-      "epoch": 0.906463478717814,
-      "grad_norm": 1.8141510486602783,
+      "epoch": 1.1150614091790563,
+      "grad_norm": 1.456865906715393,
       "learning_rate": 3.444444444444445e-06,
-      "loss": 0.2412,
+      "loss": 0.1122,
       "step": 3450
     },
     {
-      "epoch": 0.9130320546505517,
-      "grad_norm": 1.7190760374069214,
+      "epoch": 1.123141564318035,
+      "grad_norm": 1.0266293287277222,
       "learning_rate": 3.3888888888888893e-06,
-      "loss": 0.217,
+      "loss": 0.0972,
       "step": 3475
     },
     {
-      "epoch": 0.9196006305832896,
-      "grad_norm": 1.7403192520141602,
+      "epoch": 1.1312217194570136,
+      "grad_norm": 1.4198247194290161,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.2386,
+      "loss": 0.11,
       "step": 3500
     },
     {
-      "epoch": 0.9261692065160273,
-      "grad_norm": 1.5826787948608398,
+      "epoch": 1.1393018745959922,
+      "grad_norm": 1.3642793893814087,
       "learning_rate": 3.277777777777778e-06,
-      "loss": 0.1965,
+      "loss": 0.1031,
       "step": 3525
     },
     {
-      "epoch": 0.9327377824487652,
-      "grad_norm": 1.8713033199310303,
+      "epoch": 1.1473820297349708,
+      "grad_norm": 1.474433422088623,
       "learning_rate": 3.2222222222222227e-06,
-      "loss": 0.2205,
+      "loss": 0.0999,
       "step": 3550
     },
     {
-      "epoch": 0.9393063583815029,
-      "grad_norm": 1.6279629468917847,
+      "epoch": 1.1554621848739495,
+      "grad_norm": 1.726651668548584,
       "learning_rate": 3.1666666666666667e-06,
-      "loss": 0.2031,
+      "loss": 0.1052,
       "step": 3575
     },
     {
-      "epoch": 0.9458749343142406,
-      "grad_norm": 1.5993796586990356,
+      "epoch": 1.1635423400129283,
+      "grad_norm": 1.0963325500488281,
       "learning_rate": 3.1111111111111116e-06,
-      "loss": 0.218,
+      "loss": 0.0889,
       "step": 3600
     },
     {
-      "epoch": 0.9524435102469785,
-      "grad_norm": 1.8748666048049927,
+      "epoch": 1.171622495151907,
+      "grad_norm": 1.0337573289871216,
       "learning_rate": 3.055555555555556e-06,
-      "loss": 0.2198,
+      "loss": 0.1078,
       "step": 3625
     },
     {
-      "epoch": 0.9590120861797162,
-      "grad_norm": 2.0783231258392334,
+      "epoch": 1.1797026502908856,
+      "grad_norm": 1.2517417669296265,
       "learning_rate": 3e-06,
-      "loss": 0.2285,
+      "loss": 0.0983,
       "step": 3650
     },
     {
-      "epoch": 0.965580662112454,
-      "grad_norm": 1.6366628408432007,
+      "epoch": 1.1877828054298643,
+      "grad_norm": 1.1968861818313599,
       "learning_rate": 2.944444444444445e-06,
-      "loss": 0.2244,
+      "loss": 0.0981,
       "step": 3675
     },
     {
-      "epoch": 0.9721492380451918,
-      "grad_norm": 1.4321212768554688,
+      "epoch": 1.195862960568843,
+      "grad_norm": 1.646301031112671,
       "learning_rate": 2.888888888888889e-06,
-      "loss": 0.2129,
+      "loss": 0.0982,
       "step": 3700
     },
     {
-      "epoch": 0.9787178139779296,
-      "grad_norm": 1.9172074794769287,
+      "epoch": 1.2039431157078215,
+      "grad_norm": 1.5741595029830933,
       "learning_rate": 2.8333333333333335e-06,
-      "loss": 0.2308,
+      "loss": 0.0954,
       "step": 3725
     },
     {
-      "epoch": 0.9852863899106674,
-      "grad_norm": 1.601163387298584,
+      "epoch": 1.2120232708468002,
+      "grad_norm": 0.865523099899292,
       "learning_rate": 2.7777777777777783e-06,
-      "loss": 0.2249,
+      "loss": 0.0936,
       "step": 3750
     },
     {
-      "epoch": 0.9918549658434052,
-      "grad_norm": 1.920453429222107,
+      "epoch": 1.220103425985779,
+      "grad_norm": 1.109647512435913,
       "learning_rate": 2.7222222222222224e-06,
-      "loss": 0.241,
+      "loss": 0.1092,
       "step": 3775
     },
     {
-      "epoch": 0.9984235417761429,
-      "grad_norm": 1.421399712562561,
+      "epoch": 1.2281835811247577,
+      "grad_norm": 1.2372595071792603,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 0.245,
+      "loss": 0.1021,
       "step": 3800
     },
     {
-      "epoch": 1.0049921177088808,
-      "grad_norm": 1.2747637033462524,
+      "epoch": 1.2362637362637363,
+      "grad_norm": 1.5520275831222534,
       "learning_rate": 2.6111111111111113e-06,
-      "loss": 0.183,
+      "loss": 0.0972,
       "step": 3825
     },
     {
-      "epoch": 1.0115606936416186,
-      "grad_norm": 1.276963472366333,
+      "epoch": 1.244343891402715,
+      "grad_norm": 1.1914920806884766,
       "learning_rate": 2.5555555555555557e-06,
-      "loss": 0.1669,
+      "loss": 0.0945,
       "step": 3850
     },
     {
-      "epoch": 1.0181292695743562,
-      "grad_norm": 1.6604379415512085,
+      "epoch": 1.2524240465416936,
+      "grad_norm": 1.269389033317566,
       "learning_rate": 2.5e-06,
-      "loss": 0.1685,
+      "loss": 0.0949,
       "step": 3875
     },
     {
-      "epoch": 1.024697845507094,
-      "grad_norm": 1.4884883165359497,
+      "epoch": 1.2605042016806722,
+      "grad_norm": 1.683188557624817,
       "learning_rate": 2.4444444444444447e-06,
-      "loss": 0.1837,
+      "loss": 0.0923,
       "step": 3900
     },
     {
-      "epoch": 1.031266421439832,
-      "grad_norm": 1.2239917516708374,
+      "epoch": 1.2685843568196509,
+      "grad_norm": 1.4076889753341675,
       "learning_rate": 2.388888888888889e-06,
-      "loss": 0.161,
+      "loss": 0.0992,
       "step": 3925
     },
     {
-      "epoch": 1.0378349973725696,
-      "grad_norm": 1.7419662475585938,
+      "epoch": 1.2766645119586295,
+      "grad_norm": 1.0736989974975586,
       "learning_rate": 2.3333333333333336e-06,
-      "loss": 0.1903,
+      "loss": 0.0927,
       "step": 3950
     },
     {
-      "epoch": 1.0444035733053074,
-      "grad_norm": 1.5548486709594727,
+      "epoch": 1.2847446670976082,
+      "grad_norm": 0.8500351905822754,
       "learning_rate": 2.277777777777778e-06,
-      "loss": 0.159,
+      "loss": 0.092,
       "step": 3975
     },
     {
-      "epoch": 1.0509721492380453,
-      "grad_norm": 1.7805577516555786,
+      "epoch": 1.292824822236587,
+      "grad_norm": 1.2514774799346924,
       "learning_rate": 2.222222222222222e-06,
-      "loss": 0.1779,
+      "loss": 0.099,
       "step": 4000
     },
     {
-      "epoch": 1.0509721492380453,
-      "eval_loss": 0.21787120401859283,
-      "eval_runtime": 531.6511,
-      "eval_samples_per_second": 2.295,
-      "eval_steps_per_second": 0.145,
-      "eval_wer": 0.16243625065947595,
+      "epoch": 1.292824822236587,
+      "eval_loss": 0.1433822065591812,
+      "eval_runtime": 422.4096,
+      "eval_samples_per_second": 2.367,
+      "eval_steps_per_second": 0.149,
+      "eval_wer": 0.11287782194554864,
       "step": 4000
     },
     {
-      "epoch": 1.057540725170783,
-      "grad_norm": 1.646399974822998,
+      "epoch": 1.3009049773755657,
+      "grad_norm": 1.5008472204208374,
       "learning_rate": 2.166666666666667e-06,
-      "loss": 0.1891,
+      "loss": 0.0973,
       "step": 4025
     },
     {
-      "epoch": 1.0641093011035208,
-      "grad_norm": 2.159637928009033,
+      "epoch": 1.3089851325145443,
+      "grad_norm": 1.1695759296417236,
       "learning_rate": 2.1111111111111114e-06,
-      "loss": 0.1712,
+      "loss": 0.0903,
       "step": 4050
     },
     {
-      "epoch": 1.0706778770362586,
-      "grad_norm": 1.5517977476119995,
+      "epoch": 1.317065287653523,
+      "grad_norm": 1.221191644668579,
       "learning_rate": 2.0555555555555555e-06,
-      "loss": 0.1582,
+      "loss": 0.0955,
       "step": 4075
     },
     {
-      "epoch": 1.0772464529689962,
-      "grad_norm": 1.410233974456787,
+      "epoch": 1.3251454427925016,
+      "grad_norm": 1.3147417306900024,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.1828,
+      "loss": 0.1003,
       "step": 4100
     },
     {
-      "epoch": 1.083815028901734,
-      "grad_norm": 1.6741544008255005,
+      "epoch": 1.3332255979314802,
+      "grad_norm": 1.5507885217666626,
       "learning_rate": 1.944444444444445e-06,
-      "loss": 0.1597,
+      "loss": 0.1098,
       "step": 4125
     },
     {
-      "epoch": 1.090383604834472,
-      "grad_norm": 1.508376121520996,
+      "epoch": 1.341305753070459,
+      "grad_norm": 1.1320619583129883,
       "learning_rate": 1.888888888888889e-06,
-      "loss": 0.166,
+      "loss": 0.0902,
       "step": 4150
     },
     {
-      "epoch": 1.0969521807672096,
-      "grad_norm": 1.4631690979003906,
+      "epoch": 1.3493859082094377,
+      "grad_norm": 1.003406047821045,
       "learning_rate": 1.8333333333333333e-06,
-      "loss": 0.159,
+      "loss": 0.0887,
       "step": 4175
     },
     {
-      "epoch": 1.1035207566999474,
-      "grad_norm": 1.7129231691360474,
+      "epoch": 1.3574660633484164,
+      "grad_norm": 1.2398234605789185,
       "learning_rate": 1.777777777777778e-06,
-      "loss": 0.1682,
+      "loss": 0.1044,
       "step": 4200
     },
     {
-      "epoch": 1.1100893326326853,
-      "grad_norm": 1.7223049402236938,
+      "epoch": 1.365546218487395,
+      "grad_norm": 1.3185756206512451,
       "learning_rate": 1.7222222222222224e-06,
-      "loss": 0.1865,
+      "loss": 0.0931,
       "step": 4225
     },
     {
-      "epoch": 1.116657908565423,
-      "grad_norm": 1.8834460973739624,
+      "epoch": 1.3736263736263736,
+      "grad_norm": 1.8669284582138062,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.1753,
+      "loss": 0.0998,
       "step": 4250
     },
     {
-      "epoch": 1.1232264844981608,
-      "grad_norm": 1.6631660461425781,
+      "epoch": 1.3817065287653523,
+      "grad_norm": 1.4902080297470093,
       "learning_rate": 1.6111111111111113e-06,
-      "loss": 0.1585,
+      "loss": 0.0944,
       "step": 4275
     },
     {
-      "epoch": 1.1297950604308986,
-      "grad_norm": 1.6180058717727661,
+      "epoch": 1.389786683904331,
+      "grad_norm": 1.2797434329986572,
       "learning_rate": 1.5555555555555558e-06,
-      "loss": 0.1839,
+      "loss": 0.1002,
       "step": 4300
     },
     {
-      "epoch": 1.1363636363636362,
-      "grad_norm": 1.635799527168274,
+      "epoch": 1.3978668390433096,
+      "grad_norm": 1.3823866844177246,
       "learning_rate": 1.5e-06,
-      "loss": 0.1685,
+      "loss": 0.0848,
       "step": 4325
     },
     {
-      "epoch": 1.142932212296374,
-      "grad_norm": 1.4893510341644287,
+      "epoch": 1.4059469941822882,
+      "grad_norm": 1.5034900903701782,
       "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.1678,
+      "loss": 0.089,
       "step": 4350
     },
     {
-      "epoch": 1.149500788229112,
-      "grad_norm": 1.4803720712661743,
+      "epoch": 1.4140271493212668,
+      "grad_norm": 1.0645461082458496,
       "learning_rate": 1.3888888888888892e-06,
-      "loss": 0.1607,
+      "loss": 0.0814,
       "step": 4375
     },
     {
-      "epoch": 1.1560693641618498,
-      "grad_norm": 1.4839766025543213,
+      "epoch": 1.4221073044602457,
+      "grad_norm": 1.3759708404541016,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.187,
+      "loss": 0.084,
       "step": 4400
     },
     {
-      "epoch": 1.1626379400945874,
-      "grad_norm": 1.518723487854004,
+      "epoch": 1.4301874595992243,
+      "grad_norm": 1.6364449262619019,
       "learning_rate": 1.2777777777777779e-06,
-      "loss": 0.1628,
+      "loss": 0.1011,
       "step": 4425
     },
     {
-      "epoch": 1.1692065160273253,
-      "grad_norm": 1.740968942642212,
+      "epoch": 1.438267614738203,
+      "grad_norm": 1.8542567491531372,
       "learning_rate": 1.2222222222222223e-06,
-      "loss": 0.1521,
+      "loss": 0.1041,
       "step": 4450
     },
     {
-      "epoch": 1.1757750919600631,
-      "grad_norm": 1.941607117652893,
+      "epoch": 1.4463477698771816,
+      "grad_norm": 1.2790472507476807,
       "learning_rate": 1.1666666666666668e-06,
-      "loss": 0.1558,
+      "loss": 0.0768,
       "step": 4475
     },
     {
-      "epoch": 1.1823436678928008,
-      "grad_norm": 1.47350013256073,
+      "epoch": 1.4544279250161603,
+      "grad_norm": 1.3523430824279785,
       "learning_rate": 1.111111111111111e-06,
-      "loss": 0.1758,
+      "loss": 0.0862,
       "step": 4500
     },
     {
-      "epoch": 1.1889122438255386,
-      "grad_norm": 1.4565749168395996,
+      "epoch": 1.4625080801551391,
+      "grad_norm": 1.2620325088500977,
       "learning_rate": 1.0555555555555557e-06,
-      "loss": 0.1804,
+      "loss": 0.0931,
       "step": 4525
     },
     {
-      "epoch": 1.1954808197582765,
-      "grad_norm": 1.6238913536071777,
+      "epoch": 1.4705882352941178,
+      "grad_norm": 1.5867211818695068,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.1596,
+      "loss": 0.094,
       "step": 4550
     },
     {
-      "epoch": 1.202049395691014,
-      "grad_norm": 1.6127305030822754,
+      "epoch": 1.4786683904330964,
+      "grad_norm": 1.3185136318206787,
       "learning_rate": 9.444444444444445e-07,
-      "loss": 0.1801,
+      "loss": 0.0987,
       "step": 4575
     },
     {
-      "epoch": 1.208617971623752,
-      "grad_norm": 1.5954357385635376,
+      "epoch": 1.486748545572075,
+      "grad_norm": 1.5266519784927368,
       "learning_rate": 8.88888888888889e-07,
-      "loss": 0.1616,
+      "loss": 0.0936,
       "step": 4600
     },
     {
-      "epoch": 1.2151865475564898,
-      "grad_norm": 1.5588889122009277,
+      "epoch": 1.4948287007110537,
+      "grad_norm": 1.2671607732772827,
       "learning_rate": 8.333333333333333e-07,
-      "loss": 0.1727,
+      "loss": 0.0968,
       "step": 4625
     },
     {
-      "epoch": 1.2217551234892277,
-      "grad_norm": 1.2232043743133545,
+      "epoch": 1.5029088558500323,
+      "grad_norm": 0.9610714912414551,
       "learning_rate": 7.777777777777779e-07,
-      "loss": 0.1617,
+      "loss": 0.1071,
       "step": 4650
     },
     {
-      "epoch": 1.2283236994219653,
-      "grad_norm": 1.3071914911270142,
+      "epoch": 1.510989010989011,
+      "grad_norm": 1.585331678390503,
       "learning_rate": 7.222222222222222e-07,
-      "loss": 0.1623,
+      "loss": 0.0785,
       "step": 4675
     },
     {
-      "epoch": 1.2348922753547031,
-      "grad_norm": 1.1346869468688965,
+      "epoch": 1.5190691661279896,
+      "grad_norm": 1.5121099948883057,
       "learning_rate": 6.666666666666667e-07,
-      "loss": 0.155,
+      "loss": 0.0802,
       "step": 4700
     },
     {
-      "epoch": 1.241460851287441,
-      "grad_norm": 1.572501540184021,
+      "epoch": 1.5271493212669682,
+      "grad_norm": 1.6761611700057983,
       "learning_rate": 6.111111111111112e-07,
-      "loss": 0.1742,
+      "loss": 0.0956,
       "step": 4725
     },
     {
-      "epoch": 1.2480294272201786,
-      "grad_norm": 1.6292036771774292,
+      "epoch": 1.5352294764059469,
+      "grad_norm": 1.0066338777542114,
       "learning_rate": 5.555555555555555e-07,
-      "loss": 0.1535,
+      "loss": 0.0875,
       "step": 4750
     },
     {
-      "epoch": 1.2545980031529165,
-      "grad_norm": 1.8764897584915161,
+      "epoch": 1.5433096315449255,
+      "grad_norm": 0.9917683601379395,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.1564,
+      "loss": 0.0871,
       "step": 4775
     },
     {
-      "epoch": 1.2611665790856543,
-      "grad_norm": 1.7719995975494385,
+      "epoch": 1.5513897866839044,
+      "grad_norm": 1.4825103282928467,
       "learning_rate": 4.444444444444445e-07,
-      "loss": 0.1779,
+      "loss": 0.1008,
       "step": 4800
     },
     {
-      "epoch": 1.267735155018392,
-      "grad_norm": 1.4942494630813599,
+      "epoch": 1.559469941822883,
+      "grad_norm": 1.3838367462158203,
       "learning_rate": 3.8888888888888895e-07,
-      "loss": 0.1743,
+      "loss": 0.0903,
       "step": 4825
     },
     {
-      "epoch": 1.2743037309511298,
-      "grad_norm": 1.6207414865493774,
+      "epoch": 1.5675500969618616,
+      "grad_norm": 1.0742937326431274,
       "learning_rate": 3.3333333333333335e-07,
-      "loss": 0.1731,
+      "loss": 0.0918,
       "step": 4850
     },
     {
-      "epoch": 1.2808723068838677,
-      "grad_norm": 1.4872610569000244,
+      "epoch": 1.5756302521008403,
+      "grad_norm": 1.3103052377700806,
       "learning_rate": 2.7777777777777776e-07,
-      "loss": 0.1708,
+      "loss": 0.0964,
       "step": 4875
     },
     {
-      "epoch": 1.2874408828166053,
-      "grad_norm": 1.8898082971572876,
+      "epoch": 1.5837104072398192,
+      "grad_norm": 1.6678719520568848,
       "learning_rate": 2.2222222222222224e-07,
-      "loss": 0.1819,
+      "loss": 0.1042,
       "step": 4900
     },
     {
-      "epoch": 1.2940094587493431,
-      "grad_norm": 1.4731193780899048,
+      "epoch": 1.5917905623787978,
+      "grad_norm": 1.3988322019577026,
       "learning_rate": 1.6666666666666668e-07,
-      "loss": 0.1714,
+      "loss": 0.0925,
       "step": 4925
     },
     {
-      "epoch": 1.300578034682081,
-      "grad_norm": 1.4309362173080444,
+      "epoch": 1.5998707175177764,
+      "grad_norm": 1.1961122751235962,
       "learning_rate": 1.1111111111111112e-07,
-      "loss": 0.1668,
+      "loss": 0.0931,
       "step": 4950
     },
     {
-      "epoch": 1.3071466106148186,
-      "grad_norm": 1.6482913494110107,
+      "epoch": 1.607950872656755,
+      "grad_norm": 1.1993706226348877,
       "learning_rate": 5.555555555555556e-08,
-      "loss": 0.1928,
+      "loss": 0.1012,
       "step": 4975
     },
     {
-      "epoch": 1.3137151865475565,
-      "grad_norm": 1.3571888208389282,
+      "epoch": 1.6160310277957337,
+      "grad_norm": 1.3308823108673096,
       "learning_rate": 0.0,
-      "loss": 0.1626,
+      "loss": 0.0874,
       "step": 5000
     },
     {
-      "epoch": 1.3137151865475565,
-      "eval_loss": 0.21081987023353577,
-      "eval_runtime": 712.9205,
-      "eval_samples_per_second": 1.711,
-      "eval_steps_per_second": 0.108,
-      "eval_wer": 0.1561639017527405,
+      "epoch": 1.6160310277957337,
+      "eval_loss": 0.1364615559577942,
+      "eval_runtime": 437.3566,
+      "eval_samples_per_second": 2.286,
+      "eval_steps_per_second": 0.144,
+      "eval_wer": 0.10755268881722042,
       "step": 5000
     },
     {
-      "epoch": 1.3137151865475565,
+      "epoch": 1.6160310277957337,
       "step": 5000,
       "total_flos": 5.435725490631475e+20,
-      "train_loss": 0.2802161669254303,
-      "train_runtime": 36290.7707,
-      "train_samples_per_second": 4.409,
-      "train_steps_per_second": 0.138
+      "train_loss": 0.17819489703178407,
+      "train_runtime": 33619.5535,
+      "train_samples_per_second": 4.759,
+      "train_steps_per_second": 0.149
     }
   ],
   "logging_steps": 25,