End of training

README.md

@@ -2,6 +2,8 @@
 license: apache-2.0
 base_model: facebook/wav2vec2-large-xlsr-53
 tags:
+- automatic-speech-recognition
+- DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv
 - generated_from_trainer
 metrics:
 - wer
@@ -15,7 +17,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # wav2vec2-xlsr-53-ft-btb-ccv-cy
 
-This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on an unknown dataset.
+This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the DEWIBRYNJONES/BANC-TRAWSGRIFIADAU-BANGOR-CLEAN-WITH-CCV - DEFAULT dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.4061
 - Wer: 0.3193

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 0.5807200929152149,
-    "eval_loss": 0.5355480313301086,
-    "eval_runtime": 163.5057,
+    "epoch": 1.9357336430507162,
+    "eval_loss": 0.4060620963573456,
+    "eval_runtime": 153.2569,
     "eval_samples": 5656,
-    "eval_samples_per_second": 34.592,
-    "eval_steps_per_second": 4.324,
-    "eval_wer": 0.41856173067355684,
-    "total_flos": 3.3874766991231493e+18,
-    "train_loss": 1.2740035095214843,
-    "train_runtime": 6100.931,
+    "eval_samples_per_second": 36.905,
+    "eval_steps_per_second": 4.613,
+    "eval_wer": 0.31928551941069794,
+    "total_flos": 1.1255918428180738e+19,
+    "train_loss": 0.7262251342773437,
+    "train_runtime": 19143.2125,
     "train_samples": 41326,
-    "train_samples_per_second": 3.934,
-    "train_steps_per_second": 0.492
+    "train_samples_per_second": 4.179,
+    "train_steps_per_second": 0.522
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.5807200929152149,
-    "eval_loss": 0.5355480313301086,
-    "eval_runtime": 163.5057,
+    "epoch": 1.9357336430507162,
+    "eval_loss": 0.4060620963573456,
+    "eval_runtime": 153.2569,
     "eval_samples": 5656,
-    "eval_samples_per_second": 34.592,
-    "eval_steps_per_second": 4.324,
-    "eval_wer": 0.41856173067355684
+    "eval_samples_per_second": 36.905,
+    "eval_steps_per_second": 4.613,
+    "eval_wer": 0.31928551941069794
 }

runs/May31_11-27-51_af3b5d38d2a2/events.out.tfevents.1717170637.af3b5d38d2a2.1031.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:477c2644b43db37098819bc6a6eff044ab6d630b91462330aa510c375c75ff08
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.5807200929152149,
-    "total_flos": 3.3874766991231493e+18,
-    "train_loss": 1.2740035095214843,
-    "train_runtime": 6100.931,
+    "epoch": 1.9357336430507162,
+    "total_flos": 1.1255918428180738e+19,
+    "train_loss": 0.7262251342773437,
+    "train_runtime": 19143.2125,
     "train_samples": 41326,
-    "train_samples_per_second": 3.934,
-    "train_steps_per_second": 0.492
+    "train_samples_per_second": 4.179,
+    "train_steps_per_second": 0.522
 }

trainer_state.json

@@ -1,341 +1,1069 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.5807200929152149,
+  "epoch": 1.9357336430507162,
   "eval_steps": 100,
-  "global_step": 3000,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.019357336430507164,
-      "eval_loss": 3.5493686199188232,
-      "eval_runtime": 162.4972,
-      "eval_samples_per_second": 34.807,
-      "eval_steps_per_second": 4.351,
+      "eval_loss": 3.5565404891967773,
+      "eval_runtime": 151.5266,
+      "eval_samples_per_second": 37.327,
+      "eval_steps_per_second": 4.666,
       "eval_wer": 1.0,
       "step": 100
     },
     {
       "epoch": 0.03871467286101433,
-      "eval_loss": 3.0426220893859863,
-      "eval_runtime": 160.0022,
-      "eval_samples_per_second": 35.35,
-      "eval_steps_per_second": 4.419,
+      "eval_loss": 3.0301756858825684,
+      "eval_runtime": 150.582,
+      "eval_samples_per_second": 37.561,
+      "eval_steps_per_second": 4.695,
       "eval_wer": 1.0,
       "step": 200
     },
     {
       "epoch": 0.05807200929152149,
-      "eval_loss": 2.8965415954589844,
-      "eval_runtime": 160.2187,
-      "eval_samples_per_second": 35.302,
-      "eval_steps_per_second": 4.413,
+      "eval_loss": 2.9460911750793457,
+      "eval_runtime": 148.9065,
+      "eval_samples_per_second": 37.984,
+      "eval_steps_per_second": 4.748,
       "eval_wer": 1.0,
       "step": 300
     },
     {
       "epoch": 0.07742934572202866,
-      "eval_loss": 1.826250433921814,
-      "eval_runtime": 161.1204,
-      "eval_samples_per_second": 35.104,
-      "eval_steps_per_second": 4.388,
-      "eval_wer": 0.9828762176822712,
+      "eval_loss": 1.8142520189285278,
+      "eval_runtime": 149.8655,
+      "eval_samples_per_second": 37.741,
+      "eval_steps_per_second": 4.718,
+      "eval_wer": 0.940732775914365,
       "step": 400
     },
     {
       "epoch": 0.09678668215253582,
-      "grad_norm": 4.890473365783691,
-      "learning_rate": 0.0002982,
-      "loss": 3.9715,
+      "grad_norm": 3.132490396499634,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 3.9521,
       "step": 500
     },
     {
       "epoch": 0.09678668215253582,
-      "eval_loss": 1.3860080242156982,
-      "eval_runtime": 161.4467,
-      "eval_samples_per_second": 35.033,
-      "eval_steps_per_second": 4.379,
-      "eval_wer": 0.8748856542183563,
+      "eval_loss": 1.4195518493652344,
+      "eval_runtime": 150.5171,
+      "eval_samples_per_second": 37.577,
+      "eval_steps_per_second": 4.697,
+      "eval_wer": 0.8693007655149171,
       "step": 500
     },
     {
       "epoch": 0.11614401858304298,
-      "eval_loss": 1.308407187461853,
-      "eval_runtime": 161.0068,
-      "eval_samples_per_second": 35.129,
-      "eval_steps_per_second": 4.391,
-      "eval_wer": 0.8153456051098522,
+      "eval_loss": 1.16689133644104,
+      "eval_runtime": 150.5387,
+      "eval_samples_per_second": 37.572,
+      "eval_steps_per_second": 4.696,
+      "eval_wer": 0.8055239042865626,
       "step": 600
     },
     {
       "epoch": 0.13550135501355012,
-      "eval_loss": 1.0549893379211426,
-      "eval_runtime": 161.8736,
-      "eval_samples_per_second": 34.941,
-      "eval_steps_per_second": 4.368,
-      "eval_wer": 0.7336906806181894,
+      "eval_loss": 1.0756505727767944,
+      "eval_runtime": 151.2385,
+      "eval_samples_per_second": 37.398,
+      "eval_steps_per_second": 4.675,
+      "eval_wer": 0.7596251063215163,
       "step": 700
     },
     {
       "epoch": 0.1548586914440573,
-      "eval_loss": 1.0011754035949707,
-      "eval_runtime": 161.7881,
-      "eval_samples_per_second": 34.959,
-      "eval_steps_per_second": 4.37,
-      "eval_wer": 0.7190062749755259,
+      "eval_loss": 0.9944618344306946,
+      "eval_runtime": 151.1646,
+      "eval_samples_per_second": 37.416,
+      "eval_steps_per_second": 4.677,
+      "eval_wer": 0.7223925149652549,
       "step": 800
     },
     {
       "epoch": 0.17421602787456447,
-      "eval_loss": 0.913667619228363,
-      "eval_runtime": 162.0369,
-      "eval_samples_per_second": 34.906,
-      "eval_steps_per_second": 4.363,
-      "eval_wer": 0.6752258830704049,
+      "eval_loss": 0.9381263256072998,
+      "eval_runtime": 151.6289,
+      "eval_samples_per_second": 37.302,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.6870857472998347,
       "step": 900
     },
     {
       "epoch": 0.19357336430507163,
-      "grad_norm": 3.201047658920288,
-      "learning_rate": 0.0002406,
-      "loss": 1.0155,
+      "grad_norm": 7.335289001464844,
+      "learning_rate": 0.0002844,
+      "loss": 1.0266,
       "step": 1000
     },
     {
       "epoch": 0.19357336430507163,
-      "eval_loss": 0.8486206531524658,
-      "eval_runtime": 161.6295,
-      "eval_samples_per_second": 34.994,
-      "eval_steps_per_second": 4.374,
-      "eval_wer": 0.6469483718765547,
+      "eval_loss": 0.8977694511413574,
+      "eval_runtime": 156.0202,
+      "eval_samples_per_second": 36.252,
+      "eval_steps_per_second": 4.531,
+      "eval_wer": 0.661472292211648,
       "step": 1000
     },
     {
       "epoch": 0.2129307007355788,
-      "eval_loss": 0.8534524440765381,
-      "eval_runtime": 161.4191,
-      "eval_samples_per_second": 35.039,
-      "eval_steps_per_second": 4.38,
-      "eval_wer": 0.6111761968191812,
+      "eval_loss": 0.8770694136619568,
+      "eval_runtime": 151.6589,
+      "eval_samples_per_second": 37.294,
+      "eval_steps_per_second": 4.662,
+      "eval_wer": 0.6450385967164706,
       "step": 1100
     },
     {
       "epoch": 0.23228803716608595,
-      "eval_loss": 0.8349705934524536,
-      "eval_runtime": 162.6145,
-      "eval_samples_per_second": 34.782,
-      "eval_steps_per_second": 4.348,
-      "eval_wer": 0.6192807048514708,
+      "eval_loss": 0.851553201675415,
+      "eval_runtime": 151.5945,
+      "eval_samples_per_second": 37.31,
+      "eval_steps_per_second": 4.664,
+      "eval_wer": 0.640432668389209,
       "step": 1200
     },
     {
       "epoch": 0.2516453735965931,
-      "eval_loss": 0.7680675983428955,
-      "eval_runtime": 161.7995,
-      "eval_samples_per_second": 34.957,
-      "eval_steps_per_second": 4.37,
-      "eval_wer": 0.5669785431143779,
+      "eval_loss": 0.8273979425430298,
+      "eval_runtime": 151.4524,
+      "eval_samples_per_second": 37.345,
+      "eval_steps_per_second": 4.668,
+      "eval_wer": 0.6138081558633307,
       "step": 1300
     },
     {
       "epoch": 0.27100271002710025,
-      "eval_loss": 0.7377049326896667,
-      "eval_runtime": 162.5136,
-      "eval_samples_per_second": 34.803,
-      "eval_steps_per_second": 4.35,
-      "eval_wer": 0.5559211054227985,
+      "eval_loss": 0.7992698550224304,
+      "eval_runtime": 152.8076,
+      "eval_samples_per_second": 37.014,
+      "eval_steps_per_second": 4.627,
+      "eval_wer": 0.596973247099228,
       "step": 1400
     },
     {
       "epoch": 0.29036004645760743,
-      "grad_norm": 5.448112487792969,
-      "learning_rate": 0.00018059999999999997,
-      "loss": 0.7987,
+      "grad_norm": 4.0737223625183105,
+      "learning_rate": 0.00026861052631578947,
+      "loss": 0.8454,
       "step": 1500
     },
     {
       "epoch": 0.29036004645760743,
-      "eval_loss": 0.7129804491996765,
-      "eval_runtime": 162.7944,
-      "eval_samples_per_second": 34.743,
-      "eval_steps_per_second": 4.343,
-      "eval_wer": 0.5437242220474715,
+      "eval_loss": 0.7768516540527344,
+      "eval_runtime": 152.3743,
+      "eval_samples_per_second": 37.119,
+      "eval_steps_per_second": 4.64,
+      "eval_wer": 0.5887563993516394,
       "step": 1500
     },
     {
       "epoch": 0.3097173828881146,
-      "eval_loss": 0.7039781808853149,
-      "eval_runtime": 162.4512,
-      "eval_samples_per_second": 34.817,
-      "eval_steps_per_second": 4.352,
-      "eval_wer": 0.5451846383463594,
+      "eval_loss": 0.7664207220077515,
+      "eval_runtime": 154.3668,
+      "eval_samples_per_second": 36.64,
+      "eval_steps_per_second": 4.58,
+      "eval_wer": 0.5997977885124617,
       "step": 1600
     },
     {
       "epoch": 0.32907471931862176,
-      "eval_loss": 0.6728500127792358,
-      "eval_runtime": 166.2937,
-      "eval_samples_per_second": 34.012,
-      "eval_steps_per_second": 4.252,
-      "eval_wer": 0.5050954085153504,
+      "eval_loss": 0.7400562763214111,
+      "eval_runtime": 153.7228,
+      "eval_samples_per_second": 36.793,
+      "eval_steps_per_second": 4.599,
+      "eval_wer": 0.5592110542279854,
       "step": 1700
     },
     {
       "epoch": 0.34843205574912894,
-      "eval_loss": 0.6646420359611511,
-      "eval_runtime": 164.3307,
-      "eval_samples_per_second": 34.418,
-      "eval_steps_per_second": 4.302,
-      "eval_wer": 0.511338286979827,
+      "eval_loss": 0.746478796005249,
+      "eval_runtime": 151.7535,
+      "eval_samples_per_second": 37.271,
+      "eval_steps_per_second": 4.659,
+      "eval_wer": 0.5650206223620228,
       "step": 1800
     },
     {
       "epoch": 0.3677893921796361,
-      "eval_loss": 0.6530969142913818,
-      "eval_runtime": 163.5666,
-      "eval_samples_per_second": 34.579,
-      "eval_steps_per_second": 4.322,
-      "eval_wer": 0.49691065782927574,
+      "eval_loss": 0.7252949476242065,
+      "eval_runtime": 151.7548,
+      "eval_samples_per_second": 37.271,
+      "eval_steps_per_second": 4.659,
+      "eval_wer": 0.5791272808974338,
       "step": 1900
     },
     {
       "epoch": 0.38714672861014326,
-      "grad_norm": 2.6550886631011963,
-      "learning_rate": 0.00012059999999999999,
-      "loss": 0.6851,
+      "grad_norm": 2.4802448749542236,
+      "learning_rate": 0.0002528210526315789,
+      "loss": 0.7537,
       "step": 2000
     },
     {
       "epoch": 0.38714672861014326,
-      "eval_loss": 0.6413969397544861,
-      "eval_runtime": 162.7473,
-      "eval_samples_per_second": 34.753,
-      "eval_steps_per_second": 4.344,
-      "eval_wer": 0.5037954775240326,
+      "eval_loss": 0.7039346098899841,
+      "eval_runtime": 152.7969,
+      "eval_samples_per_second": 37.016,
+      "eval_steps_per_second": 4.627,
+      "eval_wer": 0.5343518800853782,
       "step": 2000
     },
     {
       "epoch": 0.4065040650406504,
-      "eval_loss": 0.6108531355857849,
-      "eval_runtime": 163.3073,
-      "eval_samples_per_second": 34.634,
-      "eval_steps_per_second": 4.329,
-      "eval_wer": 0.467654186259248,
+      "eval_loss": 0.6932350397109985,
+      "eval_runtime": 152.4406,
+      "eval_samples_per_second": 37.103,
+      "eval_steps_per_second": 4.638,
+      "eval_wer": 0.5168429330294811,
       "step": 2100
     },
     {
       "epoch": 0.4258614014711576,
-      "eval_loss": 0.6034538745880127,
-      "eval_runtime": 164.2548,
-      "eval_samples_per_second": 34.434,
-      "eval_steps_per_second": 4.304,
-      "eval_wer": 0.46924299080419185,
+      "eval_loss": 0.696869432926178,
+      "eval_runtime": 153.0527,
+      "eval_samples_per_second": 36.955,
+      "eval_steps_per_second": 4.619,
+      "eval_wer": 0.5364381890837894,
       "step": 2200
     },
     {
       "epoch": 0.4452187379016647,
-      "eval_loss": 0.5801683664321899,
-      "eval_runtime": 163.2622,
-      "eval_samples_per_second": 34.644,
-      "eval_steps_per_second": 4.33,
-      "eval_wer": 0.458955882588949,
+      "eval_loss": 0.6781283617019653,
+      "eval_runtime": 152.1378,
+      "eval_samples_per_second": 37.177,
+      "eval_steps_per_second": 4.647,
+      "eval_wer": 0.5173725345444624,
       "step": 2300
     },
     {
       "epoch": 0.4645760743321719,
-      "eval_loss": 0.571967363357544,
-      "eval_runtime": 164.105,
-      "eval_samples_per_second": 34.466,
-      "eval_steps_per_second": 4.308,
-      "eval_wer": 0.44545906822230424,
+      "eval_loss": 0.6760829091072083,
+      "eval_runtime": 151.9712,
+      "eval_samples_per_second": 37.218,
+      "eval_steps_per_second": 4.652,
+      "eval_wer": 0.5050312143923223,
       "step": 2400
     },
     {
       "epoch": 0.48393341076267904,
-      "grad_norm": 6.736985683441162,
-      "learning_rate": 6.0599999999999996e-05,
-      "loss": 0.5979,
+      "grad_norm": 3.791292667388916,
+      "learning_rate": 0.0002370315789473684,
+      "loss": 0.681,
       "step": 2500
     },
     {
       "epoch": 0.48393341076267904,
-      "eval_loss": 0.569513201713562,
-      "eval_runtime": 163.6068,
-      "eval_samples_per_second": 34.571,
-      "eval_steps_per_second": 4.321,
-      "eval_wer": 0.4425542841552856,
+      "eval_loss": 0.6720712780952454,
+      "eval_runtime": 152.2414,
+      "eval_samples_per_second": 37.152,
+      "eval_steps_per_second": 4.644,
+      "eval_wer": 0.528718845789668,
       "step": 2500
     },
     {
       "epoch": 0.5032907471931862,
-      "eval_loss": 0.5556703209877014,
-      "eval_runtime": 163.3416,
-      "eval_samples_per_second": 34.627,
-      "eval_steps_per_second": 4.328,
-      "eval_wer": 0.43513986294554735,
+      "eval_loss": 0.6598270535469055,
+      "eval_runtime": 151.7192,
+      "eval_samples_per_second": 37.279,
+      "eval_steps_per_second": 4.66,
+      "eval_wer": 0.5195069891351447,
       "step": 2600
     },
     {
       "epoch": 0.5226480836236934,
-      "eval_loss": 0.5499459505081177,
-      "eval_runtime": 163.0587,
-      "eval_samples_per_second": 34.687,
-      "eval_steps_per_second": 4.336,
-      "eval_wer": 0.4269711607902297,
+      "eval_loss": 0.6555168628692627,
+      "eval_runtime": 152.5678,
+      "eval_samples_per_second": 37.072,
+      "eval_steps_per_second": 4.634,
+      "eval_wer": 0.4975846961210701,
       "step": 2700
     },
     {
       "epoch": 0.5420054200542005,
-      "eval_loss": 0.5451160073280334,
-      "eval_runtime": 163.5278,
-      "eval_samples_per_second": 34.587,
-      "eval_steps_per_second": 4.323,
-      "eval_wer": 0.425815666575725,
+      "eval_loss": 0.6535276770591736,
+      "eval_runtime": 152.5246,
+      "eval_samples_per_second": 37.083,
+      "eval_steps_per_second": 4.635,
+      "eval_wer": 0.49936608303509816,
       "step": 2800
     },
     {
       "epoch": 0.5613627564847077,
-      "eval_loss": 0.5382982492446899,
-      "eval_runtime": 163.1776,
-      "eval_samples_per_second": 34.662,
-      "eval_steps_per_second": 4.333,
-      "eval_wer": 0.42172329123268765,
+      "eval_loss": 0.6258506178855896,
+      "eval_runtime": 151.843,
+      "eval_samples_per_second": 37.249,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.48192133010222915,
       "step": 2900
     },
     {
       "epoch": 0.5807200929152149,
-      "grad_norm": 4.4167070388793945,
-      "learning_rate": 6e-07,
-      "loss": 0.5753,
+      "grad_norm": 9.4619779586792,
+      "learning_rate": 0.00022124210526315786,
+      "loss": 0.6737,
       "step": 3000
     },
     {
       "epoch": 0.5807200929152149,
-      "eval_loss": 0.5355480313301086,
-      "eval_runtime": 164.5272,
-      "eval_samples_per_second": 34.377,
-      "eval_steps_per_second": 4.297,
-      "eval_wer": 0.41856173067355684,
+      "eval_loss": 0.629943311214447,
+      "eval_runtime": 151.8389,
+      "eval_samples_per_second": 37.25,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.48022018584198617,
       "step": 3000
     },
     {
-      "epoch": 0.5807200929152149,
-      "step": 3000,
-      "total_flos": 3.3874766991231493e+18,
-      "train_loss": 1.2740035095214843,
-      "train_runtime": 6100.931,
-      "train_samples_per_second": 3.934,
-      "train_steps_per_second": 0.492
+      "epoch": 0.6000774293457221,
+      "eval_loss": 0.6378594636917114,
+      "eval_runtime": 151.6255,
+      "eval_samples_per_second": 37.302,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.4893197027812104,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6194347657762292,
+      "eval_loss": 0.6225672364234924,
+      "eval_runtime": 153.0144,
+      "eval_samples_per_second": 36.964,
+      "eval_steps_per_second": 4.62,
+      "eval_wer": 0.4806053505801544,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6387921022067363,
+      "eval_loss": 0.6088670492172241,
+      "eval_runtime": 152.2222,
+      "eval_samples_per_second": 37.156,
+      "eval_steps_per_second": 4.645,
+      "eval_wer": 0.4627112387860891,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6581494386372435,
+      "eval_loss": 0.6028585433959961,
+      "eval_runtime": 153.0615,
+      "eval_samples_per_second": 36.952,
+      "eval_steps_per_second": 4.619,
+      "eval_wer": 0.47354399704707034,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6775067750677507,
+      "grad_norm": 3.4705822467803955,
+      "learning_rate": 0.00020545263157894736,
+      "loss": 0.6419,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6775067750677507,
+      "eval_loss": 0.5871421694755554,
+      "eval_runtime": 152.5739,
+      "eval_samples_per_second": 37.071,
+      "eval_steps_per_second": 4.634,
+      "eval_wer": 0.4592126590810611,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6968641114982579,
+      "eval_loss": 0.6001027226448059,
+      "eval_runtime": 152.1697,
+      "eval_samples_per_second": 37.169,
+      "eval_steps_per_second": 4.646,
+      "eval_wer": 0.4610742886488742,
+      "step": 3600
+    },
+    {
+      "epoch": 0.716221447928765,
+      "eval_loss": 0.5848923921585083,
+      "eval_runtime": 152.6563,
+      "eval_samples_per_second": 37.051,
+      "eval_steps_per_second": 4.631,
+      "eval_wer": 0.4472565036670893,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7355787843592722,
+      "eval_loss": 0.5923960208892822,
+      "eval_runtime": 152.6559,
+      "eval_samples_per_second": 37.051,
+      "eval_steps_per_second": 4.631,
+      "eval_wer": 0.46377044181605176,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7549361207897793,
+      "eval_loss": 0.5767965316772461,
+      "eval_runtime": 152.1652,
+      "eval_samples_per_second": 37.17,
+      "eval_steps_per_second": 4.646,
+      "eval_wer": 0.4584904751969957,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7742934572202865,
+      "grad_norm": 3.628082275390625,
+      "learning_rate": 0.00018966315789473683,
+      "loss": 0.6183,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7742934572202865,
+      "eval_loss": 0.5672534704208374,
+      "eval_runtime": 152.4329,
+      "eval_samples_per_second": 37.105,
+      "eval_steps_per_second": 4.638,
+      "eval_wer": 0.44531463144549116,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7936507936507936,
+      "eval_loss": 0.5575382113456726,
+      "eval_runtime": 152.2388,
+      "eval_samples_per_second": 37.152,
+      "eval_steps_per_second": 4.644,
+      "eval_wer": 0.4451862431994351,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8130081300813008,
+      "eval_loss": 0.5631808042526245,
+      "eval_runtime": 152.7545,
+      "eval_samples_per_second": 37.027,
+      "eval_steps_per_second": 4.628,
+      "eval_wer": 0.4474972316284444,
+      "step": 4200
+    },
+    {
+      "epoch": 0.832365466511808,
+      "eval_loss": 0.5498641729354858,
+      "eval_runtime": 153.7788,
+      "eval_samples_per_second": 36.78,
+      "eval_steps_per_second": 4.598,
+      "eval_wer": 0.44008281041870617,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8517228029423152,
+      "eval_loss": 0.5662574172019958,
+      "eval_runtime": 152.5034,
+      "eval_samples_per_second": 37.088,
+      "eval_steps_per_second": 4.636,
+      "eval_wer": 0.43101539054099597,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8710801393728222,
+      "grad_norm": 2.376349925994873,
+      "learning_rate": 0.0001738736842105263,
+      "loss": 0.5877,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8710801393728222,
+      "eval_loss": 0.5584732294082642,
+      "eval_runtime": 152.1714,
+      "eval_samples_per_second": 37.169,
+      "eval_steps_per_second": 4.646,
+      "eval_wer": 0.4317215258943044,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8904374758033294,
+      "eval_loss": 0.5463821291923523,
+      "eval_runtime": 152.4923,
+      "eval_samples_per_second": 37.09,
+      "eval_steps_per_second": 4.636,
+      "eval_wer": 0.41997400138017366,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9097948122338366,
+      "eval_loss": 0.5381494164466858,
+      "eval_runtime": 153.2139,
+      "eval_samples_per_second": 36.916,
+      "eval_steps_per_second": 4.614,
+      "eval_wer": 0.4192197204345942,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9291521486643438,
+      "eval_loss": 0.5453722476959229,
+      "eval_runtime": 151.9737,
+      "eval_samples_per_second": 37.217,
+      "eval_steps_per_second": 4.652,
+      "eval_wer": 0.4201986808107718,
+      "step": 4800
+    },
+    {
+      "epoch": 0.948509485094851,
+      "eval_loss": 0.5237515568733215,
+      "eval_runtime": 151.8558,
+      "eval_samples_per_second": 37.246,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.41241514339362234,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9678668215253581,
+      "grad_norm": 2.5489518642425537,
+      "learning_rate": 0.0001581157894736842,
+      "loss": 0.5621,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9678668215253581,
+      "eval_loss": 0.5303541421890259,
+      "eval_runtime": 152.515,
+      "eval_samples_per_second": 37.085,
+      "eval_steps_per_second": 4.636,
+      "eval_wer": 0.41353854054661293,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9872241579558653,
+      "eval_loss": 0.5163344740867615,
+      "eval_runtime": 156.7945,
+      "eval_samples_per_second": 36.073,
+      "eval_steps_per_second": 4.509,
+      "eval_wer": 0.4061080708061177,
+      "step": 5100
+    },
+    {
+      "epoch": 1.0065814943863725,
+      "eval_loss": 0.51596599817276,
+      "eval_runtime": 153.2891,
+      "eval_samples_per_second": 36.898,
+      "eval_steps_per_second": 4.612,
+      "eval_wer": 0.39927139670363176,
+      "step": 5200
+    },
+    {
+      "epoch": 1.0259388308168795,
+      "eval_loss": 0.5088583827018738,
+      "eval_runtime": 152.7112,
+      "eval_samples_per_second": 37.037,
+      "eval_steps_per_second": 4.63,
+      "eval_wer": 0.3898509091492674,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0452961672473868,
+      "eval_loss": 0.5110610723495483,
+      "eval_runtime": 152.5555,
+      "eval_samples_per_second": 37.075,
+      "eval_steps_per_second": 4.634,
+      "eval_wer": 0.3985652613503234,
+      "step": 5400
+    },
+    {
+      "epoch": 1.064653503677894,
+      "grad_norm": 1.1362248659133911,
+      "learning_rate": 0.0001423578947368421,
+      "loss": 0.4882,
+      "step": 5500
+    },
+    {
+      "epoch": 1.064653503677894,
+      "eval_loss": 0.5010027885437012,
+      "eval_runtime": 152.1249,
+      "eval_samples_per_second": 37.18,
+      "eval_steps_per_second": 4.647,
+      "eval_wer": 0.38574248527547306,
+      "step": 5500
+    },
+    {
+      "epoch": 1.084010840108401,
+      "eval_loss": 0.49406561255455017,
+      "eval_runtime": 151.5623,
+      "eval_samples_per_second": 37.318,
+      "eval_steps_per_second": 4.665,
+      "eval_wer": 0.3858548249907721,
+      "step": 5600
+    },
+    {
+      "epoch": 1.1033681765389083,
+      "eval_loss": 0.49403733015060425,
+      "eval_runtime": 152.7631,
+      "eval_samples_per_second": 37.025,
+      "eval_steps_per_second": 4.628,
+      "eval_wer": 0.3813451878480525,
+      "step": 5700
+    },
+    {
+      "epoch": 1.1227255129694154,
+      "eval_loss": 0.4913772642612457,
+      "eval_runtime": 152.1406,
+      "eval_samples_per_second": 37.176,
+      "eval_steps_per_second": 4.647,
+      "eval_wer": 0.37815153022740766,
+      "step": 5800
+    },
+    {
+      "epoch": 1.1420828493999227,
+      "eval_loss": 0.48747047781944275,
+      "eval_runtime": 151.3195,
+      "eval_samples_per_second": 37.378,
+      "eval_steps_per_second": 4.672,
+      "eval_wer": 0.3745406108070806,
+      "step": 5900
+    },
+    {
+      "epoch": 1.1614401858304297,
+      "grad_norm": 1.0150744915008545,
+      "learning_rate": 0.00012656842105263156,
+      "loss": 0.4569,
+      "step": 6000
+    },
+    {
+      "epoch": 1.1614401858304297,
+      "eval_loss": 0.4841971695423126,
+      "eval_runtime": 151.8567,
+      "eval_samples_per_second": 37.246,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.38071929514852915,
+      "step": 6000
+    },
+    {
+      "epoch": 1.1807975222609368,
+      "eval_loss": 0.48611822724342346,
+      "eval_runtime": 150.971,
+      "eval_samples_per_second": 37.464,
+      "eval_steps_per_second": 4.683,
+      "eval_wer": 0.37370608720771614,
+      "step": 6100
+    },
+    {
+      "epoch": 1.2001548586914441,
+      "eval_loss": 0.48144644498825073,
+      "eval_runtime": 151.4548,
+      "eval_samples_per_second": 37.344,
+      "eval_steps_per_second": 4.668,
+      "eval_wer": 0.3760973182905105,
+      "step": 6200
+    },
+    {
+      "epoch": 1.2195121951219512,
+      "eval_loss": 0.47813892364501953,
+      "eval_runtime": 151.1935,
+      "eval_samples_per_second": 37.409,
+      "eval_steps_per_second": 4.676,
+      "eval_wer": 0.37409125194588433,
+      "step": 6300
+    },
+    {
+      "epoch": 1.2388695315524583,
+      "eval_loss": 0.4771001935005188,
+      "eval_runtime": 151.1732,
+      "eval_samples_per_second": 37.414,
+      "eval_steps_per_second": 4.677,
+      "eval_wer": 0.36815329556579096,
+      "step": 6400
+    },
+    {
+      "epoch": 1.2582268679829656,
+      "grad_norm": 1.3292571306228638,
+      "learning_rate": 0.00011077894736842105,
+      "loss": 0.4416,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2582268679829656,
+      "eval_loss": 0.47095027565956116,
+      "eval_runtime": 151.5037,
+      "eval_samples_per_second": 37.332,
+      "eval_steps_per_second": 4.667,
+      "eval_wer": 0.37338511659257595,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2775842044134726,
+      "eval_loss": 0.47211408615112305,
+      "eval_runtime": 150.9455,
+      "eval_samples_per_second": 37.47,
+      "eval_steps_per_second": 4.684,
+      "eval_wer": 0.3659706953828377,
+      "step": 6600
+    },
+    {
+      "epoch": 1.29694154084398,
+      "eval_loss": 0.4679400622844696,
+      "eval_runtime": 151.4191,
+      "eval_samples_per_second": 37.353,
+      "eval_steps_per_second": 4.669,
+      "eval_wer": 0.3638843863844265,
+      "step": 6700
+    },
+    {
+      "epoch": 1.316298877274487,
+      "eval_loss": 0.46228036284446716,
+      "eval_runtime": 151.3839,
+      "eval_samples_per_second": 37.362,
+      "eval_steps_per_second": 4.67,
+      "eval_wer": 0.366532393959333,
+      "step": 6800
+    },
+    {
+      "epoch": 1.3356562137049943,
+      "eval_loss": 0.46108925342559814,
+      "eval_runtime": 151.8163,
+      "eval_samples_per_second": 37.256,
+      "eval_steps_per_second": 4.657,
+      "eval_wer": 0.3601771757795574,
+      "step": 6900
+    },
+    {
+      "epoch": 1.3550135501355014,
+      "grad_norm": 0.8062695860862732,
+      "learning_rate": 9.498947368421052e-05,
+      "loss": 0.4324,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3550135501355014,
+      "eval_loss": 0.46888086199760437,
+      "eval_runtime": 152.4379,
+      "eval_samples_per_second": 37.104,
+      "eval_steps_per_second": 4.638,
+      "eval_wer": 0.3609314567251368,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3743708865660085,
+      "eval_loss": 0.4573034346103668,
+      "eval_runtime": 151.3077,
+      "eval_samples_per_second": 37.381,
+      "eval_steps_per_second": 4.673,
+      "eval_wer": 0.3602574184333424,
+      "step": 7100
+    },
+    {
+      "epoch": 1.3937282229965158,
+      "eval_loss": 0.45749789476394653,
+      "eval_runtime": 151.5824,
+      "eval_samples_per_second": 37.313,
+      "eval_steps_per_second": 4.664,
+      "eval_wer": 0.3546083356068752,
+      "step": 7200
+    },
+    {
+      "epoch": 1.4130855594270229,
+      "eval_loss": 0.4555954933166504,
+      "eval_runtime": 151.6035,
+      "eval_samples_per_second": 37.308,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.35836369180401534,
+      "step": 7300
+    },
+    {
+      "epoch": 1.43244289585753,
+      "eval_loss": 0.4495578408241272,
+      "eval_runtime": 152.5621,
+      "eval_samples_per_second": 37.073,
+      "eval_steps_per_second": 4.634,
+      "eval_wer": 0.350724591163679,
+      "step": 7400
+    },
+    {
+      "epoch": 1.4518002322880372,
+      "grad_norm": 0.7916799187660217,
+      "learning_rate": 7.92e-05,
+      "loss": 0.4255,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4518002322880372,
+      "eval_loss": 0.44609567523002625,
+      "eval_runtime": 151.8498,
+      "eval_samples_per_second": 37.247,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.34671245847442667,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4711575687185443,
+      "eval_loss": 0.44341230392456055,
+      "eval_runtime": 152.528,
+      "eval_samples_per_second": 37.082,
+      "eval_steps_per_second": 4.635,
+      "eval_wer": 0.3462470510824734,
+      "step": 7600
+    },
+    {
+      "epoch": 1.4905149051490514,
+      "eval_loss": 0.44362780451774597,
+      "eval_runtime": 152.5253,
+      "eval_samples_per_second": 37.082,
+      "eval_steps_per_second": 4.635,
+      "eval_wer": 0.3516393574168285,
+      "step": 7700
+    },
+    {
+      "epoch": 1.5098722415795587,
+      "eval_loss": 0.4406072199344635,
+      "eval_runtime": 152.4039,
+      "eval_samples_per_second": 37.112,
+      "eval_steps_per_second": 4.639,
+      "eval_wer": 0.34579769222127715,
+      "step": 7800
+    },
+    {
+      "epoch": 1.5292295780100658,
+      "eval_loss": 0.43874725699424744,
+      "eval_runtime": 152.6604,
+      "eval_samples_per_second": 37.05,
+      "eval_steps_per_second": 4.631,
+      "eval_wer": 0.3439360626534641,
+      "step": 7900
+    },
+    {
+      "epoch": 1.5485869144405728,
+      "grad_norm": 0.7491864562034607,
+      "learning_rate": 6.344210526315788e-05,
+      "loss": 0.4094,
+      "step": 8000
+    },
+    {
+      "epoch": 1.5485869144405728,
+      "eval_loss": 0.43253499269485474,
+      "eval_runtime": 153.8006,
+      "eval_samples_per_second": 36.775,
+      "eval_steps_per_second": 4.597,
+      "eval_wer": 0.3409831329941744,
+      "step": 8000
+    },
+    {
+      "epoch": 1.5679442508710801,
+      "eval_loss": 0.4359830617904663,
+      "eval_runtime": 153.3674,
+      "eval_samples_per_second": 36.879,
+      "eval_steps_per_second": 4.61,
+      "eval_wer": 0.3419299963088379,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5873015873015874,
+      "eval_loss": 0.4285949170589447,
+      "eval_runtime": 153.3711,
+      "eval_samples_per_second": 36.878,
+      "eval_steps_per_second": 4.61,
+      "eval_wer": 0.3377252812505015,
+      "step": 8200
+    },
+    {
+      "epoch": 1.6066589237320945,
+      "eval_loss": 0.43007034063339233,
+      "eval_runtime": 152.2201,
+      "eval_samples_per_second": 37.157,
+      "eval_steps_per_second": 4.645,
+      "eval_wer": 0.3335526632536791,
+      "step": 8300
+    },
+    {
+      "epoch": 1.6260162601626016,
+      "eval_loss": 0.42966797947883606,
+      "eval_runtime": 152.0163,
+      "eval_samples_per_second": 37.207,
+      "eval_steps_per_second": 4.651,
+      "eval_wer": 0.3322848293238754,
+      "step": 8400
+    },
+    {
+      "epoch": 1.645373596593109,
+      "grad_norm": 1.047472596168518,
+      "learning_rate": 4.765263157894736e-05,
+      "loss": 0.4018,
+      "step": 8500
+    },
+    {
+      "epoch": 1.645373596593109,
+      "eval_loss": 0.4270441234111786,
+      "eval_runtime": 152.8058,
+      "eval_samples_per_second": 37.014,
+      "eval_steps_per_second": 4.627,
+      "eval_wer": 0.3338575853380623,
+      "step": 8500
+    },
+    {
+      "epoch": 1.664730933023616,
+      "eval_loss": 0.4267289638519287,
+      "eval_runtime": 152.5032,
+      "eval_samples_per_second": 37.088,
+      "eval_steps_per_second": 4.636,
+      "eval_wer": 0.3319959557702492,
+      "step": 8600
+    },
+    {
+      "epoch": 1.684088269454123,
+      "eval_loss": 0.4224300980567932,
+      "eval_runtime": 152.5862,
+      "eval_samples_per_second": 37.068,
+      "eval_steps_per_second": 4.633,
+      "eval_wer": 0.33275023671582865,
+      "step": 8700
+    },
+    {
+      "epoch": 1.7034456058846303,
+      "eval_loss": 0.4207303822040558,
+      "eval_runtime": 154.5205,
+      "eval_samples_per_second": 36.604,
+      "eval_steps_per_second": 4.575,
+      "eval_wer": 0.32984545264881,
+      "step": 8800
+    },
+    {
+      "epoch": 1.7228029423151374,
+      "eval_loss": 0.4197385013103485,
+      "eval_runtime": 152.0624,
+      "eval_samples_per_second": 37.195,
+      "eval_steps_per_second": 4.649,
+      "eval_wer": 0.32978125852578194,
+      "step": 8900
+    },
+    {
+      "epoch": 1.7421602787456445,
+      "grad_norm": 1.4507739543914795,
+      "learning_rate": 3.189473684210526e-05,
+      "loss": 0.3899,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7421602787456445,
+      "eval_loss": 0.4183507561683655,
+      "eval_runtime": 157.4278,
+      "eval_samples_per_second": 35.928,
+      "eval_steps_per_second": 4.491,
+      "eval_wer": 0.3258493684903147,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7615176151761518,
+      "eval_loss": 0.4164830148220062,
+      "eval_runtime": 153.0475,
+      "eval_samples_per_second": 36.956,
+      "eval_steps_per_second": 4.619,
+      "eval_wer": 0.3262024361669689,
+      "step": 9100
+    },
+    {
+      "epoch": 1.7808749516066589,
+      "eval_loss": 0.41182050108909607,
+      "eval_runtime": 152.4839,
+      "eval_samples_per_second": 37.092,
+      "eval_steps_per_second": 4.637,
+      "eval_wer": 0.322864341769511,
+      "step": 9200
+    },
+    {
+      "epoch": 1.800232288037166,
+      "eval_loss": 0.4134317636489868,
+      "eval_runtime": 152.6353,
+      "eval_samples_per_second": 37.056,
+      "eval_steps_per_second": 4.632,
+      "eval_wer": 0.3232334579769222,
+      "step": 9300
+    },
+    {
+      "epoch": 1.8195896244676733,
+      "eval_loss": 0.4126824736595154,
+      "eval_runtime": 152.5246,
+      "eval_samples_per_second": 37.083,
+      "eval_steps_per_second": 4.635,
+      "eval_wer": 0.3209064210171559,
+      "step": 9400
+    },
+    {
+      "epoch": 1.8389469608981805,
+      "grad_norm": 1.0012460947036743,
+      "learning_rate": 1.6105263157894736e-05,
+      "loss": 0.3665,
+      "step": 9500
+    },
+    {
+      "epoch": 1.8389469608981805,
+      "eval_loss": 0.41083237528800964,
+      "eval_runtime": 152.9993,
+      "eval_samples_per_second": 36.967,
+      "eval_steps_per_second": 4.621,
+      "eval_wer": 0.32109900338624,
+      "step": 9500
+    },
+    {
+      "epoch": 1.8583042973286876,
+      "eval_loss": 0.4090138077735901,
+      "eval_runtime": 152.5291,
+      "eval_samples_per_second": 37.081,
+      "eval_steps_per_second": 4.635,
+      "eval_wer": 0.3199114121102213,
+      "step": 9600
+    },
+    {
+      "epoch": 1.8776616337591947,
+      "eval_loss": 0.407578706741333,
+      "eval_runtime": 153.0711,
+      "eval_samples_per_second": 36.95,
+      "eval_steps_per_second": 4.619,
+      "eval_wer": 0.32087432395564186,
+      "step": 9700
+    },
+    {
+      "epoch": 1.897018970189702,
+      "eval_loss": 0.40649694204330444,
+      "eval_runtime": 154.4136,
+      "eval_samples_per_second": 36.629,
+      "eval_steps_per_second": 4.579,
+      "eval_wer": 0.31981512092567926,
+      "step": 9800
+    },
+    {
+      "epoch": 1.916376306620209,
+      "eval_loss": 0.40620651841163635,
+      "eval_runtime": 153.7508,
+      "eval_samples_per_second": 36.787,
+      "eval_steps_per_second": 4.598,
+      "eval_wer": 0.31923737381842693,
+      "step": 9900
+    },
+    {
+      "epoch": 1.9357336430507162,
+      "grad_norm": 0.7244949340820312,
+      "learning_rate": 3.157894736842105e-07,
+      "loss": 0.3698,
+      "step": 10000
+    },
+    {
+      "epoch": 1.9357336430507162,
+      "eval_loss": 0.4060620963573456,
+      "eval_runtime": 153.976,
+      "eval_samples_per_second": 36.733,
+      "eval_steps_per_second": 4.592,
+      "eval_wer": 0.31928551941069794,
+      "step": 10000
+    },
+    {
+      "epoch": 1.9357336430507162,
+      "step": 10000,
+      "total_flos": 1.1255918428180738e+19,
+      "train_loss": 0.7262251342773437,
+      "train_runtime": 19143.2125,
+      "train_samples_per_second": 4.179,
+      "train_steps_per_second": 0.522
     }
   ],
   "logging_steps": 500,
-  "max_steps": 3000,
+  "max_steps": 10000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 2,
   "save_steps": 400,
-  "total_flos": 3.3874766991231493e+18,
+  "total_flos": 1.1255918428180738e+19,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null