Upload 13 files

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fb7750a10cf414e7cf57db59fee51e82cf10bd6c74934f4e5000135b87050316
+oid sha256:e5a9e65ddc813ff1ae693f2253104fc1adecbbf5df46b448937bccb9492735ad
 size 2490594117

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:073b9ce2badcff0059ea13062bc550a34d8961a512ca0cee6aa15940c3e9bafb
+oid sha256:1fd74eeb653891d4e14d34884c603b0c9994fdb00ea8028e5c413b78cfe559d0
 size 1262168365

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:03df94d786d198b7279e6cbae44d7c52f57ee3dc05dadfdcb8bd58655e851b41
-size 14575
+oid sha256:bc18e8a150b018c46bd46562aa5a0ee60953fb10ab966c334d972199dbfcd87e
+size 14639

scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c17afbca2619611e7e8b4bf54fd16906bc9961ed2d474309c8d35540b117170
+oid sha256:43834ac5197f140a7b873f304e5508b173a1d381b18f877ba9ad73867f38e7ad
 size 557

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e82518d5762a21ba4f0c30189155912d4bd11e2e762621ec47d40c6a19b601ba
+oid sha256:bc84086bb1f18591932e6528ee39cbfdfeefaa2127e9c49d971ac25093b03632
 size 627

trainer_state.json

@@ -1,451 +1,901 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 35.80246913580247,
-  "global_step": 2900,
+  "epoch": 49.9989417989418,
+  "global_step": 5900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.23,
-      "learning_rate": 0.0001,
-      "loss": 0.0625,
+      "epoch": 0.85,
+      "learning_rate": 0.0003,
+      "loss": 0.0549,
       "step": 100
     },
     {
-      "epoch": 1.23,
-      "eval_loss": 0.28051668405532837,
-      "eval_runtime": 198.6071,
-      "eval_samples_per_second": 16.374,
-      "eval_steps_per_second": 2.049,
-      "eval_wer": 0.2381097005406062,
+      "epoch": 0.85,
+      "eval_loss": 0.3047053813934326,
+      "eval_runtime": 213.654,
+      "eval_samples_per_second": 15.778,
+      "eval_steps_per_second": 1.975,
+      "eval_wer": 0.26644545348701826,
       "step": 100
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 9.987639060568604e-05,
-      "loss": 0.047,
+      "epoch": 1.69,
+      "learning_rate": 0.00029974554707379135,
+      "loss": 0.0916,
       "step": 200
     },
     {
-      "epoch": 2.47,
-      "eval_loss": 0.34423157572746277,
-      "eval_runtime": 147.7154,
-      "eval_samples_per_second": 22.015,
-      "eval_steps_per_second": 2.755,
-      "eval_wer": 0.2334325457085586,
+      "epoch": 1.69,
+      "eval_loss": 0.3038010597229004,
+      "eval_runtime": 151.8957,
+      "eval_samples_per_second": 22.193,
+      "eval_steps_per_second": 2.778,
+      "eval_wer": 0.26597974152986376,
       "step": 200
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 9.975278121137207e-05,
-      "loss": 0.0409,
+      "epoch": 2.54,
+      "learning_rate": 0.00029949109414758267,
+      "loss": 0.0917,
       "step": 300
     },
     {
-      "epoch": 3.7,
-      "eval_loss": 0.3597787916660309,
-      "eval_runtime": 156.092,
-      "eval_samples_per_second": 20.834,
-      "eval_steps_per_second": 2.607,
-      "eval_wer": 0.23124582396889995,
+      "epoch": 2.54,
+      "eval_loss": 0.31602439284324646,
+      "eval_runtime": 152.2031,
+      "eval_samples_per_second": 22.148,
+      "eval_steps_per_second": 2.773,
+      "eval_wer": 0.28134823611596227,
       "step": 300
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 9.96291718170581e-05,
-      "loss": 0.0413,
+      "epoch": 3.39,
+      "learning_rate": 0.00029923664122137405,
+      "loss": 0.0968,
       "step": 400
     },
     {
-      "epoch": 4.94,
-      "eval_loss": 0.35860675573349,
-      "eval_runtime": 151.5312,
-      "eval_samples_per_second": 21.461,
-      "eval_steps_per_second": 2.686,
-      "eval_wer": 0.2421794326671931,
+      "epoch": 3.39,
+      "eval_loss": 0.34380972385406494,
+      "eval_runtime": 153.4688,
+      "eval_samples_per_second": 21.965,
+      "eval_steps_per_second": 2.75,
+      "eval_wer": 0.2737222028175573,
       "step": 400
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 9.950556242274414e-05,
-      "loss": 0.0388,
+      "epoch": 4.24,
+      "learning_rate": 0.00029898218829516537,
+      "loss": 0.0977,
       "step": 500
     },
     {
-      "epoch": 6.17,
-      "eval_loss": 0.3748931586742401,
-      "eval_runtime": 154.1563,
-      "eval_samples_per_second": 21.095,
-      "eval_steps_per_second": 2.64,
-      "eval_wer": 0.23914231913988945,
+      "epoch": 4.24,
+      "eval_loss": 0.36153408885002136,
+      "eval_runtime": 166.8287,
+      "eval_samples_per_second": 20.206,
+      "eval_steps_per_second": 2.53,
+      "eval_wer": 0.2702875771335429,
       "step": 500
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 9.938195302843017e-05,
-      "loss": 0.0383,
+      "epoch": 5.08,
+      "learning_rate": 0.00029872773536895674,
+      "loss": 0.0916,
       "step": 600
     },
     {
-      "epoch": 7.41,
-      "eval_loss": 0.36109668016433716,
-      "eval_runtime": 163.2656,
-      "eval_samples_per_second": 19.918,
-      "eval_steps_per_second": 2.493,
-      "eval_wer": 0.2352548138249408,
+      "epoch": 5.08,
+      "eval_loss": 0.3286847770214081,
+      "eval_runtime": 171.9341,
+      "eval_samples_per_second": 19.606,
+      "eval_steps_per_second": 2.454,
+      "eval_wer": 0.2669693794388171,
       "step": 600
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 9.92583436341162e-05,
-      "loss": 0.0381,
+      "epoch": 5.93,
+      "learning_rate": 0.00029847328244274806,
+      "loss": 0.0943,
       "step": 700
     },
     {
-      "epoch": 8.64,
-      "eval_loss": 0.3883003294467926,
-      "eval_runtime": 164.0468,
-      "eval_samples_per_second": 19.824,
-      "eval_steps_per_second": 2.481,
-      "eval_wer": 0.23683411285913866,
+      "epoch": 5.93,
+      "eval_loss": 0.3330075442790985,
+      "eval_runtime": 152.0172,
+      "eval_samples_per_second": 22.175,
+      "eval_steps_per_second": 2.776,
+      "eval_wer": 0.2651647456048434,
       "step": 700
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 9.913473423980223e-05,
-      "loss": 0.0379,
+      "epoch": 6.78,
+      "learning_rate": 0.0002982213740458015,
+      "loss": 0.0959,
       "step": 800
     },
     {
-      "epoch": 9.88,
-      "eval_loss": 0.3676028549671173,
-      "eval_runtime": 174.6686,
-      "eval_samples_per_second": 18.618,
-      "eval_steps_per_second": 2.33,
-      "eval_wer": 0.2324606693798214,
+      "epoch": 6.78,
+      "eval_loss": 0.3155308663845062,
+      "eval_runtime": 176.3551,
+      "eval_samples_per_second": 19.115,
+      "eval_steps_per_second": 2.393,
+      "eval_wer": 0.2723250669460938,
       "step": 800
     },
     {
-      "epoch": 11.11,
-      "learning_rate": 9.901112484548825e-05,
-      "loss": 0.0364,
+      "epoch": 7.63,
+      "learning_rate": 0.00029796692111959286,
+      "loss": 0.0953,
       "step": 900
     },
     {
-      "epoch": 11.11,
-      "eval_loss": 0.37265580892562866,
-      "eval_runtime": 150.5621,
-      "eval_samples_per_second": 21.599,
-      "eval_steps_per_second": 2.703,
-      "eval_wer": 0.23331106116746644,
+      "epoch": 7.63,
+      "eval_loss": 0.3184454143047333,
+      "eval_runtime": 179.411,
+      "eval_samples_per_second": 18.789,
+      "eval_steps_per_second": 2.352,
+      "eval_wer": 0.25940156013505644,
       "step": 900
     },
     {
-      "epoch": 12.35,
-      "learning_rate": 9.88875154511743e-05,
-      "loss": 0.0355,
+      "epoch": 8.47,
+      "learning_rate": 0.0002977124681933842,
+      "loss": 0.0989,
       "step": 1000
     },
     {
-      "epoch": 12.35,
-      "eval_loss": 0.3740839958190918,
-      "eval_runtime": 180.5633,
-      "eval_samples_per_second": 18.01,
-      "eval_steps_per_second": 2.254,
-      "eval_wer": 0.23282512300309785,
+      "epoch": 8.47,
+      "eval_loss": 0.3282919228076935,
+      "eval_runtime": 177.9041,
+      "eval_samples_per_second": 18.948,
+      "eval_steps_per_second": 2.372,
+      "eval_wer": 0.28070788217487486,
       "step": 1000
     },
     {
-      "epoch": 13.58,
-      "learning_rate": 9.876390605686032e-05,
-      "loss": 0.0365,
+      "epoch": 9.32,
+      "learning_rate": 0.00029745801526717556,
+      "loss": 0.0962,
       "step": 1100
     },
     {
-      "epoch": 13.58,
-      "eval_loss": 0.37167179584503174,
-      "eval_runtime": 177.8674,
-      "eval_samples_per_second": 18.283,
-      "eval_steps_per_second": 2.288,
-      "eval_wer": 0.23580149425985544,
+      "epoch": 9.32,
+      "eval_loss": 0.3113383948802948,
+      "eval_runtime": 183.182,
+      "eval_samples_per_second": 18.402,
+      "eval_steps_per_second": 2.304,
+      "eval_wer": 0.2656886715566422,
       "step": 1100
     },
     {
-      "epoch": 14.81,
-      "learning_rate": 9.864029666254637e-05,
-      "loss": 0.0343,
+      "epoch": 10.17,
+      "learning_rate": 0.0002972035623409669,
+      "loss": 0.0911,
       "step": 1200
     },
     {
-      "epoch": 14.81,
-      "eval_loss": 0.3842860460281372,
-      "eval_runtime": 182.4399,
-      "eval_samples_per_second": 17.825,
-      "eval_steps_per_second": 2.231,
-      "eval_wer": 0.23580149425985544,
+      "epoch": 10.17,
+      "eval_loss": 0.31265875697135925,
+      "eval_runtime": 178.4394,
+      "eval_samples_per_second": 18.892,
+      "eval_steps_per_second": 2.365,
+      "eval_wer": 0.2594597741297008,
       "step": 1200
     },
     {
-      "epoch": 16.05,
-      "learning_rate": 9.851668726823239e-05,
-      "loss": 0.0358,
+      "epoch": 11.02,
+      "learning_rate": 0.00029694910941475825,
+      "loss": 0.093,
       "step": 1300
     },
     {
-      "epoch": 16.05,
-      "eval_loss": 0.3594246208667755,
-      "eval_runtime": 180.3747,
-      "eval_samples_per_second": 18.029,
-      "eval_steps_per_second": 2.256,
-      "eval_wer": 0.23616594788313186,
+      "epoch": 11.02,
+      "eval_loss": 0.33750081062316895,
+      "eval_runtime": 185.9856,
+      "eval_samples_per_second": 18.125,
+      "eval_steps_per_second": 2.269,
+      "eval_wer": 0.2635347537548027,
       "step": 1300
     },
     {
-      "epoch": 17.28,
-      "learning_rate": 9.839307787391843e-05,
-      "loss": 0.0343,
+      "epoch": 11.86,
+      "learning_rate": 0.0002966946564885496,
+      "loss": 0.0908,
       "step": 1400
     },
     {
-      "epoch": 17.28,
-      "eval_loss": 0.4051465690135956,
-      "eval_runtime": 180.0644,
-      "eval_samples_per_second": 18.06,
-      "eval_steps_per_second": 2.26,
-      "eval_wer": 0.2351940715543947,
+      "epoch": 11.86,
+      "eval_loss": 0.31224948167800903,
+      "eval_runtime": 182.1874,
+      "eval_samples_per_second": 18.503,
+      "eval_steps_per_second": 2.316,
+      "eval_wer": 0.2616136919315403,
       "step": 1400
     },
     {
-      "epoch": 18.52,
-      "learning_rate": 9.826946847960445e-05,
-      "loss": 0.0363,
+      "epoch": 12.71,
+      "learning_rate": 0.00029644020356234095,
+      "loss": 0.1039,
       "step": 1500
     },
     {
-      "epoch": 18.52,
-      "eval_loss": 0.3952256739139557,
-      "eval_runtime": 185.3946,
-      "eval_samples_per_second": 17.541,
-      "eval_steps_per_second": 2.195,
-      "eval_wer": 0.23962825730425805,
+      "epoch": 12.71,
+      "eval_loss": 0.33441564440727234,
+      "eval_runtime": 187.7233,
+      "eval_samples_per_second": 17.957,
+      "eval_steps_per_second": 2.248,
+      "eval_wer": 0.2726161369193154,
       "step": 1500
     },
     {
-      "epoch": 19.75,
-      "learning_rate": 9.814585908529048e-05,
-      "loss": 0.0382,
+      "epoch": 13.56,
+      "learning_rate": 0.00029618575063613227,
+      "loss": 0.0921,
       "step": 1600
     },
     {
-      "epoch": 19.75,
-      "eval_loss": 0.3582073450088501,
-      "eval_runtime": 179.8373,
-      "eval_samples_per_second": 18.083,
-      "eval_steps_per_second": 2.263,
-      "eval_wer": 0.23756302010569155,
+      "epoch": 13.56,
+      "eval_loss": 0.3115340769290924,
+      "eval_runtime": 189.6708,
+      "eval_samples_per_second": 17.773,
+      "eval_steps_per_second": 2.225,
+      "eval_wer": 0.26859937128885786,
       "step": 1600
     },
     {
-      "epoch": 20.99,
-      "learning_rate": 9.802224969097652e-05,
-      "loss": 0.0337,
+      "epoch": 14.41,
+      "learning_rate": 0.00029593129770992364,
+      "loss": 0.0995,
       "step": 1700
     },
     {
-      "epoch": 20.99,
-      "eval_loss": 0.3877179026603699,
-      "eval_runtime": 183.3138,
-      "eval_samples_per_second": 17.74,
-      "eval_steps_per_second": 2.22,
-      "eval_wer": 0.2347081333900261,
+      "epoch": 14.41,
+      "eval_loss": 0.3103960156440735,
+      "eval_runtime": 183.1481,
+      "eval_samples_per_second": 18.406,
+      "eval_steps_per_second": 2.304,
+      "eval_wer": 0.2650483176155548,
       "step": 1700
     },
     {
-      "epoch": 22.22,
-      "learning_rate": 9.789864029666255e-05,
-      "loss": 0.0331,
+      "epoch": 15.25,
+      "learning_rate": 0.00029567684478371497,
+      "loss": 0.1027,
       "step": 1800
     },
     {
-      "epoch": 22.22,
-      "eval_loss": 0.3826364576816559,
-      "eval_runtime": 174.2395,
-      "eval_samples_per_second": 18.664,
-      "eval_steps_per_second": 2.336,
-      "eval_wer": 0.2383526696227905,
+      "epoch": 15.25,
+      "eval_loss": 0.33657944202423096,
+      "eval_runtime": 185.2922,
+      "eval_samples_per_second": 18.193,
+      "eval_steps_per_second": 2.277,
+      "eval_wer": 0.28891605541972293,
       "step": 1800
     },
     {
-      "epoch": 23.46,
-      "learning_rate": 9.777503090234858e-05,
-      "loss": 0.0321,
+      "epoch": 16.1,
+      "learning_rate": 0.00029542239185750634,
+      "loss": 0.1001,
       "step": 1900
     },
     {
-      "epoch": 23.46,
-      "eval_loss": 0.3872096538543701,
-      "eval_runtime": 181.6396,
-      "eval_samples_per_second": 17.904,
-      "eval_steps_per_second": 2.241,
-      "eval_wer": 0.23847415416388265,
+      "epoch": 16.1,
+      "eval_loss": 0.32664933800697327,
+      "eval_runtime": 182.6597,
+      "eval_samples_per_second": 18.455,
+      "eval_steps_per_second": 2.31,
+      "eval_wer": 0.2692979392245896,
       "step": 1900
     },
     {
-      "epoch": 24.69,
-      "learning_rate": 9.765142150803462e-05,
-      "loss": 0.0342,
+      "epoch": 16.95,
+      "learning_rate": 0.0002951679389312977,
+      "loss": 0.0955,
       "step": 2000
     },
     {
-      "epoch": 24.69,
-      "eval_loss": 0.4173298180103302,
-      "eval_runtime": 185.9384,
-      "eval_samples_per_second": 17.49,
-      "eval_steps_per_second": 2.189,
-      "eval_wer": 0.24266537083156167,
+      "epoch": 16.95,
+      "eval_loss": 0.32146599888801575,
+      "eval_runtime": 175.25,
+      "eval_samples_per_second": 19.235,
+      "eval_steps_per_second": 2.408,
+      "eval_wer": 0.25986727209221094,
       "step": 2000
     },
     {
-      "epoch": 25.93,
-      "learning_rate": 9.752781211372065e-05,
-      "loss": 0.0348,
+      "epoch": 17.8,
+      "learning_rate": 0.00029491348600508904,
+      "loss": 0.0872,
       "step": 2100
     },
     {
-      "epoch": 25.93,
-      "eval_loss": 0.36838769912719727,
-      "eval_runtime": 184.1892,
-      "eval_samples_per_second": 17.656,
-      "eval_steps_per_second": 2.21,
-      "eval_wer": 0.23671262831804654,
+      "epoch": 17.8,
+      "eval_loss": 0.31995928287506104,
+      "eval_runtime": 168.2812,
+      "eval_samples_per_second": 20.032,
+      "eval_steps_per_second": 2.508,
+      "eval_wer": 0.2623704738619164,
       "step": 2100
     },
     {
-      "epoch": 27.16,
-      "learning_rate": 9.740420271940668e-05,
-      "loss": 0.0332,
+      "epoch": 18.64,
+      "learning_rate": 0.0002946590330788804,
+      "loss": 0.0919,
       "step": 2200
     },
     {
-      "epoch": 27.16,
-      "eval_loss": 0.3941015899181366,
-      "eval_runtime": 184.3624,
-      "eval_samples_per_second": 17.639,
-      "eval_steps_per_second": 2.208,
-      "eval_wer": 0.23233918483872928,
+      "epoch": 18.64,
+      "eval_loss": 0.3285907208919525,
+      "eval_runtime": 176.4154,
+      "eval_samples_per_second": 19.108,
+      "eval_steps_per_second": 2.392,
+      "eval_wer": 0.26405867970660146,
       "step": 2200
     },
     {
-      "epoch": 28.4,
-      "learning_rate": 9.728059332509271e-05,
-      "loss": 0.0339,
+      "epoch": 19.49,
+      "learning_rate": 0.00029440458015267173,
+      "loss": 0.0953,
       "step": 2300
     },
     {
-      "epoch": 28.4,
-      "eval_loss": 0.3854130506515503,
-      "eval_runtime": 186.7325,
-      "eval_samples_per_second": 17.415,
-      "eval_steps_per_second": 2.18,
-      "eval_wer": 0.24254388629046955,
+      "epoch": 19.49,
+      "eval_loss": 0.35332390666007996,
+      "eval_runtime": 173.8594,
+      "eval_samples_per_second": 19.389,
+      "eval_steps_per_second": 2.427,
+      "eval_wer": 0.2674350913959716,
       "step": 2300
     },
     {
-      "epoch": 29.63,
-      "learning_rate": 9.715698393077875e-05,
-      "loss": 0.0349,
+      "epoch": 20.34,
+      "learning_rate": 0.0002941501272264631,
+      "loss": 0.0923,
       "step": 2400
     },
     {
-      "epoch": 29.63,
-      "eval_loss": 0.39330288767814636,
-      "eval_runtime": 183.7314,
-      "eval_samples_per_second": 17.7,
-      "eval_steps_per_second": 2.215,
-      "eval_wer": 0.23774524691732976,
+      "epoch": 20.34,
+      "eval_loss": 0.30950167775154114,
+      "eval_runtime": 173.4531,
+      "eval_samples_per_second": 19.435,
+      "eval_steps_per_second": 2.433,
+      "eval_wer": 0.2600419140761439,
       "step": 2400
     },
     {
-      "epoch": 30.86,
-      "learning_rate": 9.703337453646477e-05,
-      "loss": 0.0327,
+      "epoch": 21.19,
+      "learning_rate": 0.00029389567430025443,
+      "loss": 0.0961,
       "step": 2500
     },
     {
-      "epoch": 30.86,
-      "eval_loss": 0.3882978558540344,
-      "eval_runtime": 186.0182,
-      "eval_samples_per_second": 17.482,
-      "eval_steps_per_second": 2.188,
-      "eval_wer": 0.2347081333900261,
+      "epoch": 21.19,
+      "eval_loss": 0.3377102315425873,
+      "eval_runtime": 183.4219,
+      "eval_samples_per_second": 18.378,
+      "eval_steps_per_second": 2.301,
+      "eval_wer": 0.255210152520666,
       "step": 2500
     },
     {
-      "epoch": 32.1,
-      "learning_rate": 9.690976514215081e-05,
-      "loss": 0.0335,
+      "epoch": 22.03,
+      "learning_rate": 0.00029364122137404575,
+      "loss": 0.0919,
       "step": 2600
     },
     {
-      "epoch": 32.1,
-      "eval_loss": 0.3765297532081604,
-      "eval_runtime": 152.737,
-      "eval_samples_per_second": 21.292,
-      "eval_steps_per_second": 2.665,
-      "eval_wer": 0.2403571645508109,
+      "epoch": 22.03,
+      "eval_loss": 0.3226545751094818,
+      "eval_runtime": 183.0312,
+      "eval_samples_per_second": 18.418,
+      "eval_steps_per_second": 2.306,
+      "eval_wer": 0.2614390499476074,
       "step": 2600
     },
     {
-      "epoch": 33.33,
-      "learning_rate": 9.678615574783683e-05,
-      "loss": 0.0331,
+      "epoch": 22.88,
+      "learning_rate": 0.00029338676844783713,
+      "loss": 0.0859,
       "step": 2700
     },
     {
-      "epoch": 33.33,
-      "eval_loss": 0.37883350253105164,
-      "eval_runtime": 171.3633,
-      "eval_samples_per_second": 18.977,
-      "eval_steps_per_second": 2.375,
-      "eval_wer": 0.23616594788313186,
+      "epoch": 22.88,
+      "eval_loss": 0.30848973989486694,
+      "eval_runtime": 176.2969,
+      "eval_samples_per_second": 19.121,
+      "eval_steps_per_second": 2.394,
+      "eval_wer": 0.25416230061706835,
       "step": 2700
     },
     {
-      "epoch": 34.57,
-      "learning_rate": 9.666254635352288e-05,
-      "loss": 0.0312,
+      "epoch": 23.73,
+      "learning_rate": 0.00029313231552162845,
+      "loss": 0.0915,
       "step": 2800
     },
     {
-      "epoch": 34.57,
-      "eval_loss": 0.3879595696926117,
-      "eval_runtime": 186.5981,
-      "eval_samples_per_second": 17.428,
-      "eval_steps_per_second": 2.181,
-      "eval_wer": 0.2289376176881492,
+      "epoch": 23.73,
+      "eval_loss": 0.3403824269771576,
+      "eval_runtime": 172.6719,
+      "eval_samples_per_second": 19.523,
+      "eval_steps_per_second": 2.444,
+      "eval_wer": 0.2610315519850972,
       "step": 2800
     },
     {
-      "epoch": 35.8,
-      "learning_rate": 9.65389369592089e-05,
-      "loss": 0.0306,
+      "epoch": 24.58,
+      "learning_rate": 0.00029288040712468187,
+      "loss": 0.0917,
       "step": 2900
     },
     {
-      "epoch": 35.8,
-      "eval_loss": 0.38653597235679626,
-      "eval_runtime": 182.3902,
-      "eval_samples_per_second": 17.83,
-      "eval_steps_per_second": 2.231,
-      "eval_wer": 0.23258215392091355,
+      "epoch": 24.58,
+      "eval_loss": 0.2996799647808075,
+      "eval_runtime": 178.4531,
+      "eval_samples_per_second": 18.89,
+      "eval_steps_per_second": 2.365,
+      "eval_wer": 0.2529980207241821,
       "step": 2900
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.00029262595419847324,
+      "loss": 0.0967,
+      "step": 3000
+    },
+    {
+      "epoch": 25.42,
+      "eval_loss": 0.3144609332084656,
+      "eval_runtime": 177.4531,
+      "eval_samples_per_second": 18.997,
+      "eval_steps_per_second": 2.378,
+      "eval_wer": 0.25555943648853185,
+      "step": 3000
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 0.0002923715012722646,
+      "loss": 0.0973,
+      "step": 3100
+    },
+    {
+      "epoch": 26.27,
+      "eval_loss": 0.3294685482978821,
+      "eval_runtime": 178.7969,
+      "eval_samples_per_second": 18.854,
+      "eval_steps_per_second": 2.36,
+      "eval_wer": 0.25940156013505644,
+      "step": 3100
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 0.00029211704834605594,
+      "loss": 0.0932,
+      "step": 3200
+    },
+    {
+      "epoch": 27.12,
+      "eval_loss": 0.3125886023044586,
+      "eval_runtime": 168.7969,
+      "eval_samples_per_second": 19.971,
+      "eval_steps_per_second": 2.5,
+      "eval_wer": 0.25684014437070674,
+      "step": 3200
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 0.0002918625954198473,
+      "loss": 0.0945,
+      "step": 3300
+    },
+    {
+      "epoch": 27.97,
+      "eval_loss": 0.3468785583972931,
+      "eval_runtime": 179.6094,
+      "eval_samples_per_second": 18.769,
+      "eval_steps_per_second": 2.35,
+      "eval_wer": 0.2523576667830947,
+      "step": 3300
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 0.00029160814249363864,
+      "loss": 0.0852,
+      "step": 3400
+    },
+    {
+      "epoch": 28.81,
+      "eval_loss": 0.31798404455184937,
+      "eval_runtime": 173.2031,
+      "eval_samples_per_second": 19.463,
+      "eval_steps_per_second": 2.436,
+      "eval_wer": 0.254104086622424,
+      "step": 3400
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 0.00029135368956743,
+      "loss": 0.0866,
+      "step": 3500
+    },
+    {
+      "epoch": 29.66,
+      "eval_loss": 0.31360727548599243,
+      "eval_runtime": 167.4063,
+      "eval_samples_per_second": 20.137,
+      "eval_steps_per_second": 2.521,
+      "eval_wer": 0.255210152520666,
+      "step": 3500
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.00029109923664122133,
+      "loss": 0.0844,
+      "step": 3600
+    },
+    {
+      "epoch": 30.51,
+      "eval_loss": 0.33361586928367615,
+      "eval_runtime": 166.3125,
+      "eval_samples_per_second": 20.269,
+      "eval_steps_per_second": 2.537,
+      "eval_wer": 0.2660379555245081,
+      "step": 3600
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 0.0002908447837150127,
+      "loss": 0.0847,
+      "step": 3700
+    },
+    {
+      "epoch": 31.36,
+      "eval_loss": 0.31821873784065247,
+      "eval_runtime": 178.8437,
+      "eval_samples_per_second": 18.849,
+      "eval_steps_per_second": 2.36,
+      "eval_wer": 0.2507858889276982,
+      "step": 3700
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 0.0002905903307888041,
+      "loss": 0.0885,
+      "step": 3800
+    },
+    {
+      "epoch": 32.2,
+      "eval_loss": 0.32577720284461975,
+      "eval_runtime": 175.0625,
+      "eval_samples_per_second": 19.256,
+      "eval_steps_per_second": 2.411,
+      "eval_wer": 0.2674933053906159,
+      "step": 3800
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 0.0002903358778625954,
+      "loss": 0.0855,
+      "step": 3900
+    },
+    {
+      "epoch": 33.05,
+      "eval_loss": 0.31836631894111633,
+      "eval_runtime": 169.0469,
+      "eval_samples_per_second": 19.941,
+      "eval_steps_per_second": 2.496,
+      "eval_wer": 0.25375480265455813,
+      "step": 3900
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 0.0002900814249363867,
+      "loss": 0.0813,
+      "step": 4000
+    },
+    {
+      "epoch": 33.9,
+      "eval_loss": 0.30345430970191956,
+      "eval_runtime": 175.0938,
+      "eval_samples_per_second": 19.253,
+      "eval_steps_per_second": 2.41,
+      "eval_wer": 0.2506112469437653,
+      "step": 4000
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 0.0002898269720101781,
+      "loss": 0.0822,
+      "step": 4100
+    },
+    {
+      "epoch": 34.74,
+      "eval_loss": 0.3159136176109314,
+      "eval_runtime": 174.1406,
+      "eval_samples_per_second": 19.358,
+      "eval_steps_per_second": 2.423,
+      "eval_wer": 0.2572476423332169,
+      "step": 4100
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.0002895725190839694,
+      "loss": 0.0849,
+      "step": 4200
+    },
+    {
+      "epoch": 35.59,
+      "eval_loss": 0.2940651774406433,
+      "eval_runtime": 173.233,
+      "eval_samples_per_second": 19.459,
+      "eval_steps_per_second": 2.436,
+      "eval_wer": 0.2512516008848527,
+      "step": 4200
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 0.0002893180661577608,
+      "loss": 0.0885,
+      "step": 4300
+    },
+    {
+      "epoch": 36.44,
+      "eval_loss": 0.32734107971191406,
+      "eval_runtime": 193.9206,
+      "eval_samples_per_second": 17.383,
+      "eval_steps_per_second": 2.176,
+      "eval_wer": 0.26423332169053443,
+      "step": 4300
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 0.0002890636132315521,
+      "loss": 0.0866,
+      "step": 4400
+    },
+    {
+      "epoch": 37.29,
+      "eval_loss": 0.33303678035736084,
+      "eval_runtime": 197.0429,
+      "eval_samples_per_second": 17.108,
+      "eval_steps_per_second": 2.142,
+      "eval_wer": 0.255966934451042,
+      "step": 4400
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0002888091603053435,
+      "loss": 0.0841,
+      "step": 4500
+    },
+    {
+      "epoch": 38.14,
+      "eval_loss": 0.32818496227264404,
+      "eval_runtime": 192.3874,
+      "eval_samples_per_second": 17.522,
+      "eval_steps_per_second": 2.193,
+      "eval_wer": 0.24997089300267786,
+      "step": 4500
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 0.0002885547073791348,
+      "loss": 0.0848,
+      "step": 4600
+    },
+    {
+      "epoch": 38.98,
+      "eval_loss": 0.32277733087539673,
+      "eval_runtime": 188.7845,
+      "eval_samples_per_second": 17.856,
+      "eval_steps_per_second": 2.235,
+      "eval_wer": 0.2605658400279427,
+      "step": 4600
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 0.0002883002544529262,
+      "loss": 0.0752,
+      "step": 4700
+    },
+    {
+      "epoch": 39.83,
+      "eval_loss": 0.3181003928184509,
+      "eval_runtime": 190.576,
+      "eval_samples_per_second": 17.688,
+      "eval_steps_per_second": 2.214,
+      "eval_wer": 0.2516008848527186,
+      "step": 4700
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 0.0002880458015267175,
+      "loss": 0.0827,
+      "step": 4800
+    },
+    {
+      "epoch": 40.68,
+      "eval_loss": 0.3244548439979553,
+      "eval_runtime": 196.2976,
+      "eval_samples_per_second": 17.173,
+      "eval_steps_per_second": 2.15,
+      "eval_wer": 0.2487483991151473,
+      "step": 4800
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 0.0002877913486005089,
+      "loss": 0.0765,
+      "step": 4900
+    },
+    {
+      "epoch": 41.52,
+      "eval_loss": 0.31394141912460327,
+      "eval_runtime": 194.1554,
+      "eval_samples_per_second": 17.362,
+      "eval_steps_per_second": 2.174,
+      "eval_wer": 0.24508091745255559,
+      "step": 4900
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 0.0002875368956743002,
+      "loss": 0.0777,
+      "step": 5000
+    },
+    {
+      "epoch": 42.37,
+      "eval_loss": 0.3149695098400116,
+      "eval_runtime": 192.9814,
+      "eval_samples_per_second": 17.468,
+      "eval_steps_per_second": 2.187,
+      "eval_wer": 0.24508091745255559,
+      "step": 5000
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 0.0002872824427480916,
+      "loss": 0.0804,
+      "step": 5100
+    },
+    {
+      "epoch": 43.22,
+      "eval_loss": 0.3207753598690033,
+      "eval_runtime": 190.3397,
+      "eval_samples_per_second": 17.71,
+      "eval_steps_per_second": 2.217,
+      "eval_wer": 0.25381301664920247,
+      "step": 5100
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 0.000287030534351145,
+      "loss": 0.0838,
+      "step": 5200
+    },
+    {
+      "epoch": 44.07,
+      "eval_loss": 0.3102128505706787,
+      "eval_runtime": 175.1719,
+      "eval_samples_per_second": 19.244,
+      "eval_steps_per_second": 2.409,
+      "eval_wer": 0.24845732914192573,
+      "step": 5200
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 0.0002867760814249364,
+      "loss": 0.0731,
+      "step": 5300
+    },
+    {
+      "epoch": 44.91,
+      "eval_loss": 0.29943132400512695,
+      "eval_runtime": 173.7343,
+      "eval_samples_per_second": 19.403,
+      "eval_steps_per_second": 2.429,
+      "eval_wer": 0.24449877750611246,
+      "step": 5300
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 0.0002865216284987277,
+      "loss": 0.0736,
+      "step": 5400
+    },
+    {
+      "epoch": 45.76,
+      "eval_loss": 0.3185470998287201,
+      "eval_runtime": 174.5938,
+      "eval_samples_per_second": 19.308,
+      "eval_steps_per_second": 2.417,
+      "eval_wer": 0.2591687041564792,
+      "step": 5400
+    },
+    {
+      "epoch": 46.61,
+      "learning_rate": 0.0002862671755725191,
+      "loss": 0.0795,
+      "step": 5500
+    },
+    {
+      "epoch": 46.61,
+      "eval_loss": 0.3023243546485901,
+      "eval_runtime": 174.2188,
+      "eval_samples_per_second": 19.349,
+      "eval_steps_per_second": 2.422,
+      "eval_wer": 0.24583769938293165,
+      "step": 5500
+    },
+    {
+      "epoch": 47.46,
+      "learning_rate": 0.0002860127226463104,
+      "loss": 0.0753,
+      "step": 5600
+    },
+    {
+      "epoch": 47.46,
+      "eval_loss": 0.32648247480392456,
+      "eval_runtime": 179.8281,
+      "eval_samples_per_second": 18.746,
+      "eval_steps_per_second": 2.347,
+      "eval_wer": 0.2464780533240191,
+      "step": 5600
+    },
+    {
+      "epoch": 48.3,
+      "learning_rate": 0.0002857582697201018,
+      "loss": 0.0716,
+      "step": 5700
+    },
+    {
+      "epoch": 48.3,
+      "eval_loss": 0.3370068073272705,
+      "eval_runtime": 190.9301,
+      "eval_samples_per_second": 17.656,
+      "eval_steps_per_second": 2.21,
+      "eval_wer": 0.24903946908836885,
+      "step": 5700
+    },
+    {
+      "epoch": 49.15,
+      "learning_rate": 0.0002855038167938931,
+      "loss": 0.074,
+      "step": 5800
+    },
+    {
+      "epoch": 49.15,
+      "eval_loss": 0.299947589635849,
+      "eval_runtime": 192.7259,
+      "eval_samples_per_second": 17.491,
+      "eval_steps_per_second": 2.19,
+      "eval_wer": 0.2431598556292933,
+      "step": 5800
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 0.00028524936386768447,
+      "loss": 0.0678,
+      "step": 5900
+    },
+    {
+      "epoch": 50.0,
+      "eval_loss": 0.32643795013427734,
+      "eval_runtime": 195.0205,
+      "eval_samples_per_second": 17.285,
+      "eval_steps_per_second": 2.164,
+      "eval_wer": 0.2549190825474444,
+      "step": 5900
     }
   ],
-  "max_steps": 81000,
+  "max_steps": 118000,
   "num_train_epochs": 1000,
-  "total_flos": 9.224758770642593e+19,
+  "total_flos": 1.3321101567847278e+20,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8601095f38d32bef09015c26566a62538d0cc825cfc7a586ccc7a2a2a4d0a2e
+oid sha256:205c3abc444d227c15d961d52c246f3bbe49484c6ec6d79719acbf83317f6c1f
 size 3323