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: 2.8074
 - Wer: 0.9983

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 2.8312570781426953,
-    "eval_loss": 0.44040822982788086,
-    "eval_runtime": 153.6561,
+    "epoch": 5.662514156285391,
+    "eval_loss": 2.807446241378784,
+    "eval_runtime": 152.9951,
     "eval_samples": 5656,
-    "eval_samples_per_second": 36.809,
-    "eval_steps_per_second": 4.601,
-    "eval_wer": 0.32349023446903435,
-    "total_flos": 4.569633767265734e+19,
-    "train_loss": 0.6167776947021484,
-    "train_runtime": 19530.6251,
+    "eval_samples_per_second": 36.968,
+    "eval_steps_per_second": 4.621,
+    "eval_wer": 0.998266758678243,
+    "total_flos": 9.138710330328565e+19,
+    "train_loss": 1.1088516311645509,
+    "train_runtime": 39423.4073,
     "train_samples": 112985,
-    "train_samples_per_second": 16.385,
-    "train_steps_per_second": 0.256
+    "train_samples_per_second": 16.234,
+    "train_steps_per_second": 0.254
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.8312570781426953,
-    "eval_loss": 0.44040822982788086,
-    "eval_runtime": 153.6561,
+    "epoch": 5.662514156285391,
+    "eval_loss": 2.807446241378784,
+    "eval_runtime": 152.9951,
     "eval_samples": 5656,
-    "eval_samples_per_second": 36.809,
-    "eval_steps_per_second": 4.601,
-    "eval_wer": 0.32349023446903435
+    "eval_samples_per_second": 36.968,
+    "eval_steps_per_second": 4.621,
+    "eval_wer": 0.998266758678243
 }

runs/May28_06-22-08_5cce808c63ba/events.out.tfevents.1716913381.5cce808c63ba.722.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c2f562d657b935478e0770e132f4a3fc213931aeca640f4420911659814b360
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.8312570781426953,
-    "total_flos": 4.569633767265734e+19,
-    "train_loss": 0.6167776947021484,
-    "train_runtime": 19530.6251,
+    "epoch": 5.662514156285391,
+    "total_flos": 9.138710330328565e+19,
+    "train_loss": 1.1088516311645509,
+    "train_runtime": 39423.4073,
     "train_samples": 112985,
-    "train_samples_per_second": 16.385,
-    "train_steps_per_second": 0.256
+    "train_samples_per_second": 16.234,
+    "train_steps_per_second": 0.254
 }

trainer_state.json

@@ -1,549 +1,1069 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.8312570781426953,
+  "epoch": 5.662514156285391,
   "eval_steps": 100,
-  "global_step": 5000,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.056625141562853906,
-      "eval_loss": 3.596607208251953,
-      "eval_runtime": 152.7562,
-      "eval_samples_per_second": 37.026,
-      "eval_steps_per_second": 4.628,
+      "eval_loss": 4.158196449279785,
+      "eval_runtime": 151.6577,
+      "eval_samples_per_second": 37.295,
+      "eval_steps_per_second": 4.662,
       "eval_wer": 1.0,
       "step": 100
     },
     {
       "epoch": 0.11325028312570781,
-      "eval_loss": 3.5807790756225586,
-      "eval_runtime": 150.6166,
-      "eval_samples_per_second": 37.552,
-      "eval_steps_per_second": 4.694,
+      "eval_loss": 3.127609968185425,
+      "eval_runtime": 148.7785,
+      "eval_samples_per_second": 38.016,
+      "eval_steps_per_second": 4.752,
       "eval_wer": 1.0,
       "step": 200
     },
     {
       "epoch": 0.16987542468856173,
-      "eval_loss": 2.264209270477295,
-      "eval_runtime": 151.9721,
-      "eval_samples_per_second": 37.217,
-      "eval_steps_per_second": 4.652,
-      "eval_wer": 0.9883969122626823,
+      "eval_loss": 3.407195568084717,
+      "eval_runtime": 152.9346,
+      "eval_samples_per_second": 36.983,
+      "eval_steps_per_second": 4.623,
+      "eval_wer": 1.0,
       "step": 300
     },
     {
       "epoch": 0.22650056625141562,
-      "eval_loss": 1.0231246948242188,
-      "eval_runtime": 152.2425,
-      "eval_samples_per_second": 37.151,
-      "eval_steps_per_second": 4.644,
-      "eval_wer": 0.7425976151883296,
+      "eval_loss": 2.1966774463653564,
+      "eval_runtime": 150.1881,
+      "eval_samples_per_second": 37.659,
+      "eval_steps_per_second": 4.707,
+      "eval_wer": 0.9898733770923271,
       "step": 400
     },
     {
       "epoch": 0.28312570781426954,
-      "grad_norm": 1.5508322715759277,
-      "learning_rate": 0.0002982,
-      "loss": 3.6826,
+      "grad_norm": 1.805786371231079,
+      "learning_rate": 0.0001491,
+      "loss": 4.6703,
       "step": 500
     },
     {
       "epoch": 0.28312570781426954,
-      "eval_loss": 0.9161245822906494,
-      "eval_runtime": 152.0295,
-      "eval_samples_per_second": 37.203,
-      "eval_steps_per_second": 4.65,
-      "eval_wer": 0.645680537946751,
+      "eval_loss": 1.0575733184814453,
+      "eval_runtime": 149.8633,
+      "eval_samples_per_second": 37.741,
+      "eval_steps_per_second": 4.718,
+      "eval_wer": 0.7174335189613391,
       "step": 500
     },
     {
       "epoch": 0.33975084937712347,
-      "eval_loss": 0.7942017316818237,
-      "eval_runtime": 151.7127,
-      "eval_samples_per_second": 37.281,
-      "eval_steps_per_second": 4.66,
-      "eval_wer": 0.5827382003177609,
+      "eval_loss": 0.8877128958702087,
+      "eval_runtime": 152.3639,
+      "eval_samples_per_second": 37.122,
+      "eval_steps_per_second": 4.64,
+      "eval_wer": 0.640464765450723,
       "step": 600
     },
     {
       "epoch": 0.39637599093997733,
-      "eval_loss": 0.7551485300064087,
-      "eval_runtime": 152.436,
-      "eval_samples_per_second": 37.104,
-      "eval_steps_per_second": 4.638,
-      "eval_wer": 0.5417021071720884,
+      "eval_loss": 0.7894217371940613,
+      "eval_runtime": 156.1616,
+      "eval_samples_per_second": 36.219,
+      "eval_steps_per_second": 4.527,
+      "eval_wer": 0.5861244403074899,
       "step": 700
     },
     {
       "epoch": 0.45300113250283125,
-      "eval_loss": 0.7242529392242432,
-      "eval_runtime": 151.7599,
-      "eval_samples_per_second": 37.269,
-      "eval_steps_per_second": 4.659,
-      "eval_wer": 0.5352345492770136,
+      "eval_loss": 0.7698957324028015,
+      "eval_runtime": 153.3515,
+      "eval_samples_per_second": 36.883,
+      "eval_steps_per_second": 4.61,
+      "eval_wer": 0.5876650992601627,
       "step": 800
     },
     {
       "epoch": 0.5096262740656852,
-      "eval_loss": 0.666481077671051,
-      "eval_runtime": 152.1768,
-      "eval_samples_per_second": 37.167,
-      "eval_steps_per_second": 4.646,
-      "eval_wer": 0.49851551090497664,
+      "eval_loss": 0.7604610919952393,
+      "eval_runtime": 153.0775,
+      "eval_samples_per_second": 36.949,
+      "eval_steps_per_second": 4.619,
+      "eval_wer": 0.5402577394039576,
       "step": 900
     },
     {
       "epoch": 0.5662514156285391,
-      "grad_norm": 1.319555401802063,
-      "learning_rate": 0.00026693333333333333,
-      "loss": 0.4714,
+      "grad_norm": 1.723030686378479,
+      "learning_rate": 0.0002988,
+      "loss": 0.5242,
       "step": 1000
     },
     {
       "epoch": 0.5662514156285391,
-      "eval_loss": 0.6402752995491028,
-      "eval_runtime": 152.9026,
-      "eval_samples_per_second": 36.991,
-      "eval_steps_per_second": 4.624,
-      "eval_wer": 0.48628653046813564,
+      "eval_loss": 0.742854654788971,
+      "eval_runtime": 152.8719,
+      "eval_samples_per_second": 36.998,
+      "eval_steps_per_second": 4.625,
+      "eval_wer": 0.5494856445892379,
       "step": 1000
     },
     {
       "epoch": 0.622876557191393,
-      "eval_loss": 0.6219765543937683,
-      "eval_runtime": 153.2975,
-      "eval_samples_per_second": 36.896,
-      "eval_steps_per_second": 4.612,
-      "eval_wer": 0.4727415705092199,
+      "eval_loss": 0.6761639714241028,
+      "eval_runtime": 152.3166,
+      "eval_samples_per_second": 37.133,
+      "eval_steps_per_second": 4.642,
+      "eval_wer": 0.5081125322976682,
       "step": 1100
     },
     {
       "epoch": 0.6795016987542469,
-      "eval_loss": 0.6100151538848877,
-      "eval_runtime": 152.7807,
-      "eval_samples_per_second": 37.02,
-      "eval_steps_per_second": 4.628,
-      "eval_wer": 0.46418770361573397,
+      "eval_loss": 0.67025226354599,
+      "eval_runtime": 153.9119,
+      "eval_samples_per_second": 36.748,
+      "eval_steps_per_second": 4.594,
+      "eval_wer": 0.5071977660445186,
       "step": 1200
     },
     {
       "epoch": 0.7361268403171007,
-      "eval_loss": 0.586254358291626,
-      "eval_runtime": 152.4833,
-      "eval_samples_per_second": 37.093,
-      "eval_steps_per_second": 4.637,
-      "eval_wer": 0.4441912342925005,
+      "eval_loss": 0.6186646223068237,
+      "eval_runtime": 155.1162,
+      "eval_samples_per_second": 36.463,
+      "eval_steps_per_second": 4.558,
+      "eval_wer": 0.4623421225786779,
       "step": 1300
     },
     {
       "epoch": 0.7927519818799547,
-      "eval_loss": 0.5667628645896912,
-      "eval_runtime": 154.0765,
-      "eval_samples_per_second": 36.709,
-      "eval_steps_per_second": 4.589,
-      "eval_wer": 0.43844586028149124,
+      "eval_loss": 0.6205167174339294,
+      "eval_runtime": 155.3219,
+      "eval_samples_per_second": 36.415,
+      "eval_steps_per_second": 4.552,
+      "eval_wer": 0.4741698897465937,
       "step": 1400
     },
     {
       "epoch": 0.8493771234428086,
-      "grad_norm": 1.6093908548355103,
-      "learning_rate": 0.00023366666666666664,
-      "loss": 0.3661,
+      "grad_norm": 1.6596413850784302,
+      "learning_rate": 0.00028349999999999995,
+      "loss": 0.4093,
       "step": 1500
     },
     {
       "epoch": 0.8493771234428086,
-      "eval_loss": 0.5753483176231384,
-      "eval_runtime": 154.2844,
-      "eval_samples_per_second": 36.66,
-      "eval_steps_per_second": 4.582,
-      "eval_wer": 0.44213702235560337,
+      "eval_loss": 0.6089454889297485,
+      "eval_runtime": 156.2384,
+      "eval_samples_per_second": 36.201,
+      "eval_steps_per_second": 4.525,
+      "eval_wer": 0.46065702684919196,
       "step": 1500
     },
     {
       "epoch": 0.9060022650056625,
-      "eval_loss": 0.5641067624092102,
-      "eval_runtime": 153.3007,
-      "eval_samples_per_second": 36.895,
-      "eval_steps_per_second": 4.612,
-      "eval_wer": 0.43111168172553804,
+      "eval_loss": 0.6079407334327698,
+      "eval_runtime": 151.5221,
+      "eval_samples_per_second": 37.328,
+      "eval_steps_per_second": 4.666,
+      "eval_wer": 0.4563720691370705,
       "step": 1600
     },
     {
       "epoch": 0.9626274065685164,
-      "eval_loss": 0.5466340780258179,
-      "eval_runtime": 154.2149,
-      "eval_samples_per_second": 36.676,
-      "eval_steps_per_second": 4.585,
-      "eval_wer": 0.41525573333761295,
+      "eval_loss": 0.5751839280128479,
+      "eval_runtime": 157.3729,
+      "eval_samples_per_second": 35.94,
+      "eval_steps_per_second": 4.493,
+      "eval_wer": 0.4487169199659771,
       "step": 1700
     },
     {
       "epoch": 1.0192525481313703,
-      "eval_loss": 0.5303572416305542,
-      "eval_runtime": 152.6129,
-      "eval_samples_per_second": 37.061,
-      "eval_steps_per_second": 4.633,
-      "eval_wer": 0.40233666607822055,
+      "eval_loss": 0.5519313812255859,
+      "eval_runtime": 157.5034,
+      "eval_samples_per_second": 35.91,
+      "eval_steps_per_second": 4.489,
+      "eval_wer": 0.41742228498980916,
       "step": 1800
     },
     {
       "epoch": 1.0758776896942241,
-      "eval_loss": 0.5279697179794312,
-      "eval_runtime": 153.0429,
-      "eval_samples_per_second": 36.957,
-      "eval_steps_per_second": 4.62,
-      "eval_wer": 0.3938469933077627,
+      "eval_loss": 0.5467554926872253,
+      "eval_runtime": 156.3131,
+      "eval_samples_per_second": 36.184,
+      "eval_steps_per_second": 4.523,
+      "eval_wer": 0.41071399913337936,
       "step": 1900
     },
     {
       "epoch": 1.1325028312570782,
-      "grad_norm": 0.6171467900276184,
-      "learning_rate": 0.00020046666666666667,
-      "loss": 0.3069,
+      "grad_norm": 0.5150347352027893,
+      "learning_rate": 0.0002669,
+      "loss": 0.3366,
       "step": 2000
     },
     {
       "epoch": 1.1325028312570782,
-      "eval_loss": 0.5219519138336182,
-      "eval_runtime": 153.2427,
-      "eval_samples_per_second": 36.909,
-      "eval_steps_per_second": 4.614,
-      "eval_wer": 0.3902200253566786,
+      "eval_loss": 0.5371935963630676,
+      "eval_runtime": 153.7972,
+      "eval_samples_per_second": 36.776,
+      "eval_steps_per_second": 4.597,
+      "eval_wer": 0.4080338944969588,
       "step": 2000
     },
     {
       "epoch": 1.189127972819932,
-      "eval_loss": 0.50642991065979,
-      "eval_runtime": 153.5923,
-      "eval_samples_per_second": 36.825,
-      "eval_steps_per_second": 4.603,
-      "eval_wer": 0.3928519844008281,
+      "eval_loss": 0.53590327501297,
+      "eval_runtime": 153.8993,
+      "eval_samples_per_second": 36.751,
+      "eval_steps_per_second": 4.594,
+      "eval_wer": 0.40715122530532327,
       "step": 2100
     },
     {
       "epoch": 1.245753114382786,
-      "eval_loss": 0.5032854676246643,
-      "eval_runtime": 152.9375,
-      "eval_samples_per_second": 36.982,
-      "eval_steps_per_second": 4.623,
-      "eval_wer": 0.38232414822423005,
+      "eval_loss": 0.530412495136261,
+      "eval_runtime": 152.2809,
+      "eval_samples_per_second": 37.142,
+      "eval_steps_per_second": 4.643,
+      "eval_wer": 0.4022885204859495,
       "step": 2200
     },
     {
       "epoch": 1.3023782559456398,
-      "eval_loss": 0.4920494556427002,
-      "eval_runtime": 153.4126,
-      "eval_samples_per_second": 36.868,
-      "eval_steps_per_second": 4.608,
-      "eval_wer": 0.37728490956652916,
+      "eval_loss": 0.5311455130577087,
+      "eval_runtime": 152.7946,
+      "eval_samples_per_second": 37.017,
+      "eval_steps_per_second": 4.627,
+      "eval_wer": 0.40111697774068783,
       "step": 2300
     },
     {
       "epoch": 1.3590033975084936,
-      "eval_loss": 0.49545198678970337,
-      "eval_runtime": 153.0386,
-      "eval_samples_per_second": 36.958,
-      "eval_steps_per_second": 4.62,
-      "eval_wer": 0.3736739901462021,
+      "eval_loss": 0.5186213254928589,
+      "eval_runtime": 152.2204,
+      "eval_samples_per_second": 37.157,
+      "eval_steps_per_second": 4.645,
+      "eval_wer": 0.38644862062878144,
       "step": 2400
     },
     {
       "epoch": 1.4156285390713477,
-      "grad_norm": 0.5370103716850281,
-      "learning_rate": 0.0001672,
-      "loss": 0.264,
+      "grad_norm": 0.7118180394172668,
+      "learning_rate": 0.00025026666666666666,
+      "loss": 0.2939,
       "step": 2500
     },
     {
       "epoch": 1.4156285390713477,
-      "eval_loss": 0.4954512119293213,
-      "eval_runtime": 152.6004,
-      "eval_samples_per_second": 37.064,
-      "eval_steps_per_second": 4.633,
-      "eval_wer": 0.3739147181075573,
+      "eval_loss": 0.5234143733978271,
+      "eval_runtime": 153.4973,
+      "eval_samples_per_second": 36.848,
+      "eval_steps_per_second": 4.606,
+      "eval_wer": 0.3934297315080804,
       "step": 2500
     },
     {
       "epoch": 1.4722536806342017,
-      "eval_loss": 0.4845733046531677,
-      "eval_runtime": 150.9099,
-      "eval_samples_per_second": 37.479,
-      "eval_steps_per_second": 4.685,
-      "eval_wer": 0.3651040747219592,
+      "eval_loss": 0.5212889909744263,
+      "eval_runtime": 153.2132,
+      "eval_samples_per_second": 36.916,
+      "eval_steps_per_second": 4.614,
+      "eval_wer": 0.39726533035900563,
       "step": 2600
     },
     {
       "epoch": 1.5288788221970555,
-      "eval_loss": 0.47606581449508667,
-      "eval_runtime": 150.159,
-      "eval_samples_per_second": 37.667,
-      "eval_steps_per_second": 4.708,
-      "eval_wer": 0.36447818202243587,
+      "eval_loss": 0.5155624151229858,
+      "eval_runtime": 153.779,
+      "eval_samples_per_second": 36.78,
+      "eval_steps_per_second": 4.598,
+      "eval_wer": 0.3876522604355571,
       "step": 2700
     },
     {
       "epoch": 1.5855039637599093,
-      "eval_loss": 0.4672282934188843,
-      "eval_runtime": 151.2978,
-      "eval_samples_per_second": 37.383,
-      "eval_steps_per_second": 4.673,
-      "eval_wer": 0.36266469804689383,
+      "eval_loss": 0.5052253007888794,
+      "eval_runtime": 153.3545,
+      "eval_samples_per_second": 36.882,
+      "eval_steps_per_second": 4.61,
+      "eval_wer": 0.3897546179647253,
       "step": 2800
     },
     {
       "epoch": 1.6421291053227633,
-      "eval_loss": 0.4644935131072998,
-      "eval_runtime": 151.1778,
-      "eval_samples_per_second": 37.413,
-      "eval_steps_per_second": 4.677,
-      "eval_wer": 0.36399672609972555,
+      "eval_loss": 0.49809539318084717,
+      "eval_runtime": 154.1154,
+      "eval_samples_per_second": 36.7,
+      "eval_steps_per_second": 4.587,
+      "eval_wer": 0.38331915713116466,
       "step": 2900
     },
     {
       "epoch": 1.6987542468856174,
-      "grad_norm": 1.252485990524292,
-      "learning_rate": 0.00013386666666666665,
-      "loss": 0.2498,
+      "grad_norm": 0.7669665217399597,
+      "learning_rate": 0.00023359999999999996,
+      "loss": 0.2838,
       "step": 3000
     },
     {
       "epoch": 1.6987542468856174,
-      "eval_loss": 0.46515557169914246,
-      "eval_runtime": 152.0904,
-      "eval_samples_per_second": 37.188,
-      "eval_steps_per_second": 4.649,
-      "eval_wer": 0.3588772447882396,
+      "eval_loss": 0.49897971749305725,
+      "eval_runtime": 152.4639,
+      "eval_samples_per_second": 37.097,
+      "eval_steps_per_second": 4.637,
+      "eval_wer": 0.380414373064146,
       "step": 3000
     },
     {
       "epoch": 1.7553793884484712,
-      "eval_loss": 0.45957261323928833,
-      "eval_runtime": 153.0761,
-      "eval_samples_per_second": 36.949,
-      "eval_steps_per_second": 4.619,
-      "eval_wer": 0.35335655020782847,
+      "eval_loss": 0.49996882677078247,
+      "eval_runtime": 151.3476,
+      "eval_samples_per_second": 37.371,
+      "eval_steps_per_second": 4.671,
+      "eval_wer": 0.38070324661777216,
       "step": 3100
     },
     {
       "epoch": 1.812004530011325,
-      "eval_loss": 0.450503408908844,
-      "eval_runtime": 153.7448,
-      "eval_samples_per_second": 36.788,
-      "eval_steps_per_second": 4.599,
-      "eval_wer": 0.35098136765579113,
+      "eval_loss": 0.49606603384017944,
+      "eval_runtime": 151.8767,
+      "eval_samples_per_second": 37.241,
+      "eval_steps_per_second": 4.655,
+      "eval_wer": 0.37513440644508994,
       "step": 3200
     },
     {
       "epoch": 1.8686296715741788,
-      "eval_loss": 0.4490334093570709,
-      "eval_runtime": 152.7992,
-      "eval_samples_per_second": 37.016,
-      "eval_steps_per_second": 4.627,
-      "eval_wer": 0.3522010559933238,
+      "eval_loss": 0.48593518137931824,
+      "eval_runtime": 150.9613,
+      "eval_samples_per_second": 37.467,
+      "eval_steps_per_second": 4.683,
+      "eval_wer": 0.3730962430389498,
       "step": 3300
     },
     {
       "epoch": 1.9252548131370328,
-      "eval_loss": 0.44600149989128113,
-      "eval_runtime": 153.6597,
-      "eval_samples_per_second": 36.809,
-      "eval_steps_per_second": 4.601,
-      "eval_wer": 0.34557301279067903,
+      "eval_loss": 0.481240451335907,
+      "eval_runtime": 151.7124,
+      "eval_samples_per_second": 37.281,
+      "eval_steps_per_second": 4.66,
+      "eval_wer": 0.3657139188907256,
       "step": 3400
     },
     {
       "epoch": 1.9818799546998869,
-      "grad_norm": 0.38641658425331116,
-      "learning_rate": 0.00010059999999999999,
-      "loss": 0.2304,
+      "grad_norm": 1.395547866821289,
+      "learning_rate": 0.00021696666666666664,
+      "loss": 0.2694,
       "step": 3500
     },
     {
       "epoch": 1.9818799546998869,
-      "eval_loss": 0.44401755928993225,
-      "eval_runtime": 152.5541,
-      "eval_samples_per_second": 37.075,
-      "eval_steps_per_second": 4.634,
-      "eval_wer": 0.3404535314791931,
+      "eval_loss": 0.47789159417152405,
+      "eval_runtime": 151.9051,
+      "eval_samples_per_second": 37.234,
+      "eval_steps_per_second": 4.654,
+      "eval_wer": 0.36197461122434244,
       "step": 3500
     },
     {
       "epoch": 2.0385050962627407,
-      "eval_loss": 0.44931888580322266,
-      "eval_runtime": 151.9569,
-      "eval_samples_per_second": 37.221,
-      "eval_steps_per_second": 4.653,
-      "eval_wer": 0.3424756463545762,
+      "eval_loss": 0.4943128526210785,
+      "eval_runtime": 152.1624,
+      "eval_samples_per_second": 37.171,
+      "eval_steps_per_second": 4.646,
+      "eval_wer": 0.36325849368490315,
       "step": 3600
     },
     {
       "epoch": 2.0951302378255945,
-      "eval_loss": 0.4386073052883148,
-      "eval_runtime": 152.2426,
-      "eval_samples_per_second": 37.151,
-      "eval_steps_per_second": 4.644,
-      "eval_wer": 0.33743640769687533,
+      "eval_loss": 0.48801928758621216,
+      "eval_runtime": 151.6345,
+      "eval_samples_per_second": 37.3,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.3677360337661087,
       "step": 3700
     },
     {
       "epoch": 2.1517553793884483,
-      "eval_loss": 0.4424333870410919,
-      "eval_runtime": 152.4948,
-      "eval_samples_per_second": 37.09,
-      "eval_steps_per_second": 4.636,
-      "eval_wer": 0.33806230039639873,
+      "eval_loss": 0.49899762868881226,
+      "eval_runtime": 151.9237,
+      "eval_samples_per_second": 37.229,
+      "eval_steps_per_second": 4.654,
+      "eval_wer": 0.3661632777519218,
       "step": 3800
     },
     {
       "epoch": 2.2083805209513026,
-      "eval_loss": 0.43851807713508606,
-      "eval_runtime": 151.5663,
-      "eval_samples_per_second": 37.317,
-      "eval_steps_per_second": 4.665,
-      "eval_wer": 0.3372598738585483,
+      "eval_loss": 0.5101335644721985,
+      "eval_runtime": 151.3463,
+      "eval_samples_per_second": 37.371,
+      "eval_steps_per_second": 4.671,
+      "eval_wer": 0.36991863394906194,
       "step": 3900
     },
     {
       "epoch": 2.2650056625141564,
-      "grad_norm": 0.3740696609020233,
-      "learning_rate": 6.739999999999998e-05,
-      "loss": 0.1947,
+      "grad_norm": 1.0641744136810303,
+      "learning_rate": 0.00020036666666666664,
+      "loss": 0.2419,
       "step": 4000
     },
     {
       "epoch": 2.2650056625141564,
-      "eval_loss": 0.4393535852432251,
-      "eval_runtime": 151.3645,
-      "eval_samples_per_second": 37.367,
-      "eval_steps_per_second": 4.671,
-      "eval_wer": 0.3354624384137632,
+      "eval_loss": 0.5392731428146362,
+      "eval_runtime": 151.0047,
+      "eval_samples_per_second": 37.456,
+      "eval_steps_per_second": 4.682,
+      "eval_wer": 0.3901718797644076,
       "step": 4000
     },
     {
       "epoch": 2.32163080407701,
-      "eval_loss": 0.4353909194469452,
-      "eval_runtime": 151.9247,
-      "eval_samples_per_second": 37.229,
-      "eval_steps_per_second": 4.654,
-      "eval_wer": 0.33242926610068846,
+      "eval_loss": 0.6454418301582336,
+      "eval_runtime": 151.5484,
+      "eval_samples_per_second": 37.321,
+      "eval_steps_per_second": 4.665,
+      "eval_wer": 0.4513328304793696,
       "step": 4100
     },
     {
       "epoch": 2.378255945639864,
-      "eval_loss": 0.4344673752784729,
-      "eval_runtime": 151.7721,
-      "eval_samples_per_second": 37.266,
-      "eval_steps_per_second": 4.658,
-      "eval_wer": 0.3324774116929595,
+      "eval_loss": 0.989225447177887,
+      "eval_runtime": 151.0915,
+      "eval_samples_per_second": 37.434,
+      "eval_steps_per_second": 4.679,
+      "eval_wer": 0.5936672497632842,
       "step": 4200
     },
     {
       "epoch": 2.434881087202718,
-      "eval_loss": 0.4342668950557709,
-      "eval_runtime": 151.7151,
-      "eval_samples_per_second": 37.28,
-      "eval_steps_per_second": 4.66,
-      "eval_wer": 0.33311935292323985,
+      "eval_loss": 0.7711612582206726,
+      "eval_runtime": 151.4711,
+      "eval_samples_per_second": 37.34,
+      "eval_steps_per_second": 4.668,
+      "eval_wer": 0.5166984962526681,
       "step": 4300
     },
     {
       "epoch": 2.491506228765572,
-      "eval_loss": 0.4516645669937134,
-      "eval_runtime": 150.7864,
-      "eval_samples_per_second": 37.51,
-      "eval_steps_per_second": 4.689,
-      "eval_wer": 0.3349328368987819,
+      "eval_loss": 0.6337701678276062,
+      "eval_runtime": 152.602,
+      "eval_samples_per_second": 37.064,
+      "eval_steps_per_second": 4.633,
+      "eval_wer": 0.47877581807385533,
       "step": 4400
     },
     {
       "epoch": 2.548131370328426,
-      "grad_norm": 1.0771362781524658,
-      "learning_rate": 3.4133333333333334e-05,
-      "loss": 0.1996,
+      "grad_norm": 1.0334250926971436,
+      "learning_rate": 0.00018373333333333332,
+      "loss": 0.46,
       "step": 4500
     },
     {
       "epoch": 2.548131370328426,
-      "eval_loss": 0.45709919929504395,
-      "eval_runtime": 151.5098,
-      "eval_samples_per_second": 37.331,
-      "eval_steps_per_second": 4.666,
-      "eval_wer": 0.3327662852465857,
+      "eval_loss": 0.5562150478363037,
+      "eval_runtime": 152.8682,
+      "eval_samples_per_second": 36.999,
+      "eval_steps_per_second": 4.625,
+      "eval_wer": 0.4155606554219961,
       "step": 4500
     },
     {
       "epoch": 2.6047565118912797,
-      "eval_loss": 0.44231659173965454,
-      "eval_runtime": 151.9331,
-      "eval_samples_per_second": 37.227,
-      "eval_steps_per_second": 4.653,
-      "eval_wer": 0.3254963008136605,
+      "eval_loss": 0.5376870036125183,
+      "eval_runtime": 151.9955,
+      "eval_samples_per_second": 37.212,
+      "eval_steps_per_second": 4.651,
+      "eval_wer": 0.3905730930333328,
       "step": 4600
     },
     {
       "epoch": 2.661381653454134,
-      "eval_loss": 0.44050487875938416,
-      "eval_runtime": 151.3798,
-      "eval_samples_per_second": 37.363,
-      "eval_steps_per_second": 4.67,
-      "eval_wer": 0.3253518640368474,
+      "eval_loss": 0.5686676502227783,
+      "eval_runtime": 151.879,
+      "eval_samples_per_second": 37.24,
+      "eval_steps_per_second": 4.655,
+      "eval_wer": 0.4008120556563047,
       "step": 4700
     },
     {
       "epoch": 2.7180067950169873,
-      "eval_loss": 0.4436578154563904,
-      "eval_runtime": 152.4074,
-      "eval_samples_per_second": 37.111,
-      "eval_steps_per_second": 4.639,
-      "eval_wer": 0.3237791080226605,
+      "eval_loss": 0.6321017742156982,
+      "eval_runtime": 152.7183,
+      "eval_samples_per_second": 37.036,
+      "eval_steps_per_second": 4.629,
+      "eval_wer": 0.42905746978864084,
       "step": 4800
     },
     {
       "epoch": 2.7746319365798415,
-      "eval_loss": 0.44009852409362793,
-      "eval_runtime": 152.2847,
-      "eval_samples_per_second": 37.141,
-      "eval_steps_per_second": 4.643,
-      "eval_wer": 0.3232334579769222,
+      "eval_loss": 0.5834416151046753,
+      "eval_runtime": 152.1246,
+      "eval_samples_per_second": 37.18,
+      "eval_steps_per_second": 4.648,
+      "eval_wer": 0.4202789234645568,
       "step": 4900
     },
     {
       "epoch": 2.8312570781426953,
-      "grad_norm": 0.8818329572677612,
-      "learning_rate": 7.999999999999999e-07,
-      "loss": 0.2024,
+      "grad_norm": 0.5010664463043213,
+      "learning_rate": 0.00016706666666666664,
+      "loss": 0.299,
       "step": 5000
     },
     {
       "epoch": 2.8312570781426953,
-      "eval_loss": 0.44040822982788086,
-      "eval_runtime": 151.6855,
-      "eval_samples_per_second": 37.288,
-      "eval_steps_per_second": 4.661,
-      "eval_wer": 0.32349023446903435,
+      "eval_loss": 0.5302273631095886,
+      "eval_runtime": 152.1036,
+      "eval_samples_per_second": 37.185,
+      "eval_steps_per_second": 4.648,
+      "eval_wer": 0.3929643241161272,
       "step": 5000
     },
     {
-      "epoch": 2.8312570781426953,
-      "step": 5000,
-      "total_flos": 4.569633767265734e+19,
-      "train_loss": 0.6167776947021484,
-      "train_runtime": 19530.6251,
-      "train_samples_per_second": 16.385,
-      "train_steps_per_second": 0.256
+      "epoch": 2.887882219705549,
+      "eval_loss": 0.5315878987312317,
+      "eval_runtime": 151.9165,
+      "eval_samples_per_second": 37.231,
+      "eval_steps_per_second": 4.654,
+      "eval_wer": 0.3860153102983422,
+      "step": 5100
+    },
+    {
+      "epoch": 2.9445073612684034,
+      "eval_loss": 0.5343597531318665,
+      "eval_runtime": 163.0683,
+      "eval_samples_per_second": 34.685,
+      "eval_steps_per_second": 4.336,
+      "eval_wer": 0.38002920832597775,
+      "step": 5200
+    },
+    {
+      "epoch": 3.001132502831257,
+      "eval_loss": 0.534857451915741,
+      "eval_runtime": 153.2271,
+      "eval_samples_per_second": 36.913,
+      "eval_steps_per_second": 4.614,
+      "eval_wer": 0.38415368073052913,
+      "step": 5300
+    },
+    {
+      "epoch": 3.057757644394111,
+      "eval_loss": 0.5775672793388367,
+      "eval_runtime": 151.3492,
+      "eval_samples_per_second": 37.371,
+      "eval_steps_per_second": 4.671,
+      "eval_wer": 0.4182728571199307,
+      "step": 5400
+    },
+    {
+      "epoch": 3.114382785956965,
+      "grad_norm": 1.614545226097107,
+      "learning_rate": 0.00015039999999999997,
+      "loss": 0.2839,
+      "step": 5500
+    },
+    {
+      "epoch": 3.114382785956965,
+      "eval_loss": 0.5882839560508728,
+      "eval_runtime": 150.3863,
+      "eval_samples_per_second": 37.61,
+      "eval_steps_per_second": 4.701,
+      "eval_wer": 0.41002391231082796,
+      "step": 5500
+    },
+    {
+      "epoch": 3.1710079275198186,
+      "eval_loss": 0.5722731947898865,
+      "eval_runtime": 154.6093,
+      "eval_samples_per_second": 36.583,
+      "eval_steps_per_second": 4.573,
+      "eval_wer": 0.4043587809536037,
+      "step": 5600
+    },
+    {
+      "epoch": 3.227633069082673,
+      "eval_loss": 0.5630057454109192,
+      "eval_runtime": 153.3174,
+      "eval_samples_per_second": 36.891,
+      "eval_steps_per_second": 4.611,
+      "eval_wer": 0.40779316653560366,
+      "step": 5700
+    },
+    {
+      "epoch": 3.2842582106455267,
+      "eval_loss": 0.5810334086418152,
+      "eval_runtime": 153.0372,
+      "eval_samples_per_second": 36.958,
+      "eval_steps_per_second": 4.62,
+      "eval_wer": 0.4191394777808092,
+      "step": 5800
+    },
+    {
+      "epoch": 3.3408833522083805,
+      "eval_loss": 0.5995615720748901,
+      "eval_runtime": 152.3384,
+      "eval_samples_per_second": 37.128,
+      "eval_steps_per_second": 4.641,
+      "eval_wer": 0.4228306398549213,
+      "step": 5900
+    },
+    {
+      "epoch": 3.3975084937712343,
+      "grad_norm": 9.246959686279297,
+      "learning_rate": 0.00013373333333333332,
+      "loss": 0.3019,
+      "step": 6000
+    },
+    {
+      "epoch": 3.3975084937712343,
+      "eval_loss": 0.5681526064872742,
+      "eval_runtime": 153.1128,
+      "eval_samples_per_second": 36.94,
+      "eval_steps_per_second": 4.618,
+      "eval_wer": 0.4015502880711271,
+      "step": 6000
+    },
+    {
+      "epoch": 3.454133635334088,
+      "eval_loss": 0.5560505390167236,
+      "eval_runtime": 152.0848,
+      "eval_samples_per_second": 37.19,
+      "eval_steps_per_second": 4.649,
+      "eval_wer": 0.40569080900643545,
+      "step": 6100
+    },
+    {
+      "epoch": 3.5107587768969424,
+      "eval_loss": 0.5905264616012573,
+      "eval_runtime": 152.0038,
+      "eval_samples_per_second": 37.21,
+      "eval_steps_per_second": 4.651,
+      "eval_wer": 0.41458169504581854,
+      "step": 6200
+    },
+    {
+      "epoch": 3.567383918459796,
+      "eval_loss": 0.5875168442726135,
+      "eval_runtime": 151.4059,
+      "eval_samples_per_second": 37.357,
+      "eval_steps_per_second": 4.67,
+      "eval_wer": 0.4190271380655101,
+      "step": 6300
+    },
+    {
+      "epoch": 3.62400906002265,
+      "eval_loss": 0.5877885818481445,
+      "eval_runtime": 150.4489,
+      "eval_samples_per_second": 37.594,
+      "eval_steps_per_second": 4.699,
+      "eval_wer": 0.44462454462293977,
+      "step": 6400
+    },
+    {
+      "epoch": 3.680634201585504,
+      "grad_norm": 1.084346890449524,
+      "learning_rate": 0.00011709999999999999,
+      "loss": 0.2944,
+      "step": 6500
+    },
+    {
+      "epoch": 3.680634201585504,
+      "eval_loss": 0.5938708782196045,
+      "eval_runtime": 150.586,
+      "eval_samples_per_second": 37.56,
+      "eval_steps_per_second": 4.695,
+      "eval_wer": 0.4403877325030893,
+      "step": 6500
+    },
+    {
+      "epoch": 3.7372593431483576,
+      "eval_loss": 0.590270459651947,
+      "eval_runtime": 150.087,
+      "eval_samples_per_second": 37.685,
+      "eval_steps_per_second": 4.711,
+      "eval_wer": 0.4183049541814447,
+      "step": 6600
+    },
+    {
+      "epoch": 3.793884484711212,
+      "eval_loss": 0.5807533264160156,
+      "eval_runtime": 149.6534,
+      "eval_samples_per_second": 37.794,
+      "eval_steps_per_second": 4.724,
+      "eval_wer": 0.4059475854985476,
+      "step": 6700
+    },
+    {
+      "epoch": 3.8505096262740657,
+      "eval_loss": 0.6154611706733704,
+      "eval_runtime": 150.8194,
+      "eval_samples_per_second": 37.502,
+      "eval_steps_per_second": 4.688,
+      "eval_wer": 0.410056009372342,
+      "step": 6800
+    },
+    {
+      "epoch": 3.9071347678369195,
+      "eval_loss": 0.7987228631973267,
+      "eval_runtime": 151.5296,
+      "eval_samples_per_second": 37.326,
+      "eval_steps_per_second": 4.666,
+      "eval_wer": 0.5822727929258077,
+      "step": 6900
+    },
+    {
+      "epoch": 3.9637599093997737,
+      "grad_norm": 9.350592613220215,
+      "learning_rate": 0.0001005,
+      "loss": 0.3918,
+      "step": 7000
+    },
+    {
+      "epoch": 3.9637599093997737,
+      "eval_loss": 0.9750258326530457,
+      "eval_runtime": 151.1967,
+      "eval_samples_per_second": 37.408,
+      "eval_steps_per_second": 4.676,
+      "eval_wer": 0.5545409317776957,
+      "step": 7000
+    },
+    {
+      "epoch": 4.020385050962627,
+      "eval_loss": 1.0540127754211426,
+      "eval_runtime": 150.2681,
+      "eval_samples_per_second": 37.639,
+      "eval_steps_per_second": 4.705,
+      "eval_wer": 0.5689043668052189,
+      "step": 7100
+    },
+    {
+      "epoch": 4.077010192525481,
+      "eval_loss": 0.6850923299789429,
+      "eval_runtime": 150.3993,
+      "eval_samples_per_second": 37.607,
+      "eval_steps_per_second": 4.701,
+      "eval_wer": 0.4396013544959959,
+      "step": 7200
+    },
+    {
+      "epoch": 4.133635334088336,
+      "eval_loss": 0.7331786155700684,
+      "eval_runtime": 151.5017,
+      "eval_samples_per_second": 37.333,
+      "eval_steps_per_second": 4.667,
+      "eval_wer": 0.49731187109820096,
+      "step": 7300
+    },
+    {
+      "epoch": 4.190260475651189,
+      "eval_loss": 0.9466453194618225,
+      "eval_runtime": 150.2633,
+      "eval_samples_per_second": 37.641,
+      "eval_steps_per_second": 4.705,
+      "eval_wer": 0.6394697565437885,
+      "step": 7400
+    },
+    {
+      "epoch": 4.246885617214043,
+      "grad_norm": 4.335416793823242,
+      "learning_rate": 8.386666666666665e-05,
+      "loss": 0.5378,
+      "step": 7500
+    },
+    {
+      "epoch": 4.246885617214043,
+      "eval_loss": 0.8257068991661072,
+      "eval_runtime": 151.0198,
+      "eval_samples_per_second": 37.452,
+      "eval_steps_per_second": 4.682,
+      "eval_wer": 0.485147084784388,
+      "step": 7500
+    },
+    {
+      "epoch": 4.303510758776897,
+      "eval_loss": 0.8490071296691895,
+      "eval_runtime": 150.818,
+      "eval_samples_per_second": 37.502,
+      "eval_steps_per_second": 4.688,
+      "eval_wer": 0.4867037922678179,
+      "step": 7600
+    },
+    {
+      "epoch": 4.360135900339751,
+      "eval_loss": 0.8716742396354675,
+      "eval_runtime": 150.797,
+      "eval_samples_per_second": 37.507,
+      "eval_steps_per_second": 4.688,
+      "eval_wer": 0.47112066890276194,
+      "step": 7700
+    },
+    {
+      "epoch": 4.416761041902605,
+      "eval_loss": 0.883883535861969,
+      "eval_runtime": 150.6157,
+      "eval_samples_per_second": 37.553,
+      "eval_steps_per_second": 4.694,
+      "eval_wer": 0.5860281491229478,
+      "step": 7800
+    },
+    {
+      "epoch": 4.4733861834654585,
+      "eval_loss": 2.911261558532715,
+      "eval_runtime": 150.8548,
+      "eval_samples_per_second": 37.493,
+      "eval_steps_per_second": 4.687,
+      "eval_wer": 1.0,
+      "step": 7900
+    },
+    {
+      "epoch": 4.530011325028313,
+      "grad_norm": 0.2805318534374237,
+      "learning_rate": 6.723333333333333e-05,
+      "loss": 1.3847,
+      "step": 8000
+    },
+    {
+      "epoch": 4.530011325028313,
+      "eval_loss": 2.8575596809387207,
+      "eval_runtime": 151.2637,
+      "eval_samples_per_second": 37.392,
+      "eval_steps_per_second": 4.674,
+      "eval_wer": 1.0,
+      "step": 8000
+    },
+    {
+      "epoch": 4.586636466591166,
+      "eval_loss": 2.83913516998291,
+      "eval_runtime": 151.4914,
+      "eval_samples_per_second": 37.335,
+      "eval_steps_per_second": 4.667,
+      "eval_wer": 1.0,
+      "step": 8100
+    },
+    {
+      "epoch": 4.64326160815402,
+      "eval_loss": 2.840644121170044,
+      "eval_runtime": 150.8361,
+      "eval_samples_per_second": 37.498,
+      "eval_steps_per_second": 4.687,
+      "eval_wer": 1.0,
+      "step": 8200
+    },
+    {
+      "epoch": 4.699886749716875,
+      "eval_loss": 2.856635093688965,
+      "eval_runtime": 150.3023,
+      "eval_samples_per_second": 37.631,
+      "eval_steps_per_second": 4.704,
+      "eval_wer": 1.0,
+      "step": 8300
+    },
+    {
+      "epoch": 4.756511891279728,
+      "eval_loss": 2.8454244136810303,
+      "eval_runtime": 150.6924,
+      "eval_samples_per_second": 37.533,
+      "eval_steps_per_second": 4.692,
+      "eval_wer": 0.999823466161673,
+      "step": 8400
+    },
+    {
+      "epoch": 4.813137032842582,
+      "grad_norm": 0.4468824863433838,
+      "learning_rate": 5.06e-05,
+      "loss": 2.8136,
+      "step": 8500
+    },
+    {
+      "epoch": 4.813137032842582,
+      "eval_loss": 2.8339831829071045,
+      "eval_runtime": 151.2056,
+      "eval_samples_per_second": 37.406,
+      "eval_steps_per_second": 4.676,
+      "eval_wer": 0.999919757346215,
+      "step": 8500
+    },
+    {
+      "epoch": 4.869762174405436,
+      "eval_loss": 2.836662530899048,
+      "eval_runtime": 152.389,
+      "eval_samples_per_second": 37.116,
+      "eval_steps_per_second": 4.639,
+      "eval_wer": 0.999935805876972,
+      "step": 8600
+    },
+    {
+      "epoch": 4.92638731596829,
+      "eval_loss": 2.833369016647339,
+      "eval_runtime": 151.3614,
+      "eval_samples_per_second": 37.368,
+      "eval_steps_per_second": 4.671,
+      "eval_wer": 0.9998876602847009,
+      "step": 8700
+    },
+    {
+      "epoch": 4.983012457531144,
+      "eval_loss": 2.8321285247802734,
+      "eval_runtime": 157.5502,
+      "eval_samples_per_second": 35.9,
+      "eval_steps_per_second": 4.487,
+      "eval_wer": 0.999935805876972,
+      "step": 8800
+    },
+    {
+      "epoch": 5.0396375990939974,
+      "eval_loss": 2.802509307861328,
+      "eval_runtime": 151.5865,
+      "eval_samples_per_second": 37.312,
+      "eval_steps_per_second": 4.664,
+      "eval_wer": 0.9982186130859719,
+      "step": 8900
+    },
+    {
+      "epoch": 5.096262740656852,
+      "grad_norm": 0.3411979377269745,
+      "learning_rate": 3.393333333333333e-05,
+      "loss": 2.8007,
+      "step": 9000
+    },
+    {
+      "epoch": 5.096262740656852,
+      "eval_loss": 2.8024423122406006,
+      "eval_runtime": 151.6682,
+      "eval_samples_per_second": 37.292,
+      "eval_steps_per_second": 4.661,
+      "eval_wer": 0.9974964292019066,
+      "step": 9000
+    },
+    {
+      "epoch": 5.152887882219706,
+      "eval_loss": 2.8043477535247803,
+      "eval_runtime": 153.1811,
+      "eval_samples_per_second": 36.924,
+      "eval_steps_per_second": 4.615,
+      "eval_wer": 0.9981062733706729,
+      "step": 9100
+    },
+    {
+      "epoch": 5.209513023782559,
+      "eval_loss": 2.810584783554077,
+      "eval_runtime": 151.6694,
+      "eval_samples_per_second": 37.292,
+      "eval_steps_per_second": 4.661,
+      "eval_wer": 0.9992457190544206,
+      "step": 9200
+    },
+    {
+      "epoch": 5.266138165345414,
+      "eval_loss": 2.8066723346710205,
+      "eval_runtime": 151.6427,
+      "eval_samples_per_second": 37.298,
+      "eval_steps_per_second": 4.662,
+      "eval_wer": 0.9993259617082056,
+      "step": 9300
+    },
+    {
+      "epoch": 5.322763306908267,
+      "eval_loss": 2.805284023284912,
+      "eval_runtime": 151.8566,
+      "eval_samples_per_second": 37.246,
+      "eval_steps_per_second": 4.656,
+      "eval_wer": 0.9985877292933832,
+      "step": 9400
+    },
+    {
+      "epoch": 5.379388448471121,
+      "grad_norm": 0.15991327166557312,
+      "learning_rate": 1.7299999999999997e-05,
+      "loss": 2.7935,
+      "step": 9500
+    },
+    {
+      "epoch": 5.379388448471121,
+      "eval_loss": 2.807739734649658,
+      "eval_runtime": 152.8905,
+      "eval_samples_per_second": 36.994,
+      "eval_steps_per_second": 4.624,
+      "eval_wer": 0.9978173998170468,
+      "step": 9500
+    },
+    {
+      "epoch": 5.436013590033975,
+      "eval_loss": 2.8082854747772217,
+      "eval_runtime": 151.7494,
+      "eval_samples_per_second": 37.272,
+      "eval_steps_per_second": 4.659,
+      "eval_wer": 0.9987161175394392,
+      "step": 9600
+    },
+    {
+      "epoch": 5.492638731596829,
+      "eval_loss": 2.8080484867095947,
+      "eval_runtime": 151.6292,
+      "eval_samples_per_second": 37.302,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.9988926513777664,
+      "step": 9700
+    },
+    {
+      "epoch": 5.549263873159683,
+      "eval_loss": 2.808607578277588,
+      "eval_runtime": 153.2376,
+      "eval_samples_per_second": 36.91,
+      "eval_steps_per_second": 4.614,
+      "eval_wer": 0.9986037778241402,
+      "step": 9800
+    },
+    {
+      "epoch": 5.605889014722536,
+      "eval_loss": 2.8079216480255127,
+      "eval_runtime": 151.6181,
+      "eval_samples_per_second": 37.304,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.998186516024458,
+      "step": 9900
+    },
+    {
+      "epoch": 5.662514156285391,
+      "grad_norm": 0.3985973000526428,
+      "learning_rate": 7e-07,
+      "loss": 2.7861,
+      "step": 10000
+    },
+    {
+      "epoch": 5.662514156285391,
+      "eval_loss": 2.807446241378784,
+      "eval_runtime": 151.2977,
+      "eval_samples_per_second": 37.383,
+      "eval_steps_per_second": 4.673,
+      "eval_wer": 0.998266758678243,
+      "step": 10000
+    },
+    {
+      "epoch": 5.662514156285391,
+      "step": 10000,
+      "total_flos": 9.138710330328565e+19,
+      "train_loss": 1.1088516311645509,
+      "train_runtime": 39423.4073,
+      "train_samples_per_second": 16.234,
+      "train_steps_per_second": 0.254
     }
   ],
   "logging_steps": 500,
-  "max_steps": 5000,
+  "max_steps": 10000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
+  "num_train_epochs": 6,
   "save_steps": 400,
-  "total_flos": 4.569633767265734e+19,
+  "total_flos": 9.138710330328565e+19,
   "train_batch_size": 64,
   "trial_name": null,
   "trial_params": null