End of training

README.md

@@ -3,11 +3,24 @@ license: apache-2.0
 base_model: openai/whisper-large-v3
 tags:
 - generated_from_trainer
+datasets:
+- DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv
 metrics:
 - wer
 model-index:
 - name: whisper-large-v3-ft-btb-ccv-cy
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv default
+      type: DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv
+      args: default
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.2910627217048187
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -15,7 +28,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-large-v3-ft-btb-ccv-cy
 
-This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on an unknown dataset.
+This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv default dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.4191
 - Wer: 0.2911

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 3.861003861003861,
-    "eval_loss": 0.447899729013443,
-    "eval_runtime": 2897.5042,
+    "epoch": 3.155569580309246,
+    "eval_loss": 0.4191373586654663,
+    "eval_runtime": 2975.4013,
     "eval_samples": 7022,
-    "eval_samples_per_second": 2.423,
-    "eval_steps_per_second": 0.152,
-    "eval_wer": 0.29472481531011024,
-    "total_flos": 5.435487665750016e+20,
-    "train_loss": 0.2695653033494949,
-    "train_runtime": 64824.1008,
-    "train_samples": 41435,
-    "train_samples_per_second": 2.468,
-    "train_steps_per_second": 0.077
+    "eval_samples_per_second": 2.36,
+    "eval_steps_per_second": 0.148,
+    "eval_wer": 0.2910627217048187,
+    "total_flos": 5.4352838158516224e+20,
+    "train_loss": 0.2719093270301819,
+    "train_runtime": 63684.9216,
+    "train_samples": 50697,
+    "train_samples_per_second": 2.512,
+    "train_steps_per_second": 0.079
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.861003861003861,
-    "eval_loss": 0.447899729013443,
-    "eval_runtime": 2897.5042,
+    "epoch": 3.155569580309246,
+    "eval_loss": 0.4191373586654663,
+    "eval_runtime": 2975.4013,
     "eval_samples": 7022,
-    "eval_samples_per_second": 2.423,
-    "eval_steps_per_second": 0.152,
-    "eval_wer": 0.29472481531011024
+    "eval_samples_per_second": 2.36,
+    "eval_steps_per_second": 0.148,
+    "eval_wer": 0.2910627217048187
 }

runs/Aug28_14-17-48_4dbbe0fe04c6/events.out.tfevents.1724918851.4dbbe0fe04c6.130.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27ab8e7b24d69be31189ea213647422b9c10b95fc98b94e4e4aed03fb1e10954
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.861003861003861,
-    "total_flos": 5.435487665750016e+20,
-    "train_loss": 0.2695653033494949,
-    "train_runtime": 64824.1008,
-    "train_samples": 41435,
-    "train_samples_per_second": 2.468,
-    "train_steps_per_second": 0.077
+    "epoch": 3.155569580309246,
+    "total_flos": 5.4352838158516224e+20,
+    "train_loss": 0.2719093270301819,
+    "train_runtime": 63684.9216,
+    "train_samples": 50697,
+    "train_samples_per_second": 2.512,
+    "train_steps_per_second": 0.079
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.861003861003861,
+  "epoch": 3.155569580309246,
   "eval_steps": 1000,
   "global_step": 5000,
   "is_hyper_param_search": false,
@@ -9,1458 +9,1458 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.019305019305019305,
-      "grad_norm": 7.365023136138916,
+      "epoch": 0.015777847901546228,
+      "grad_norm": 7.437136650085449,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 1.4936,
+      "loss": 1.3977,
       "step": 25
     },
     {
-      "epoch": 0.03861003861003861,
-      "grad_norm": 5.869426727294922,
+      "epoch": 0.031555695803092455,
+      "grad_norm": 6.577849864959717,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.2163,
+      "loss": 1.1112,
       "step": 50
     },
     {
-      "epoch": 0.05791505791505792,
-      "grad_norm": 5.041717052459717,
+      "epoch": 0.047333543704638686,
+      "grad_norm": 5.520799160003662,
       "learning_rate": 1.5e-06,
-      "loss": 0.866,
+      "loss": 0.8031,
       "step": 75
     },
     {
-      "epoch": 0.07722007722007722,
-      "grad_norm": 4.53203010559082,
+      "epoch": 0.06311139160618491,
+      "grad_norm": 5.328930854797363,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.7562,
+      "loss": 0.7451,
       "step": 100
     },
     {
-      "epoch": 0.09652509652509653,
-      "grad_norm": 5.75096321105957,
+      "epoch": 0.07888923950773115,
+      "grad_norm": 5.012049674987793,
       "learning_rate": 2.5e-06,
-      "loss": 0.6816,
+      "loss": 0.682,
       "step": 125
     },
     {
-      "epoch": 0.11583011583011583,
-      "grad_norm": 5.387002944946289,
+      "epoch": 0.09466708740927737,
+      "grad_norm": 4.783524036407471,
       "learning_rate": 3e-06,
-      "loss": 0.7012,
+      "loss": 0.6447,
       "step": 150
     },
     {
-      "epoch": 0.13513513513513514,
-      "grad_norm": 5.036501884460449,
+      "epoch": 0.1104449353108236,
+      "grad_norm": 4.749491214752197,
       "learning_rate": 3.5e-06,
-      "loss": 0.6297,
+      "loss": 0.6121,
       "step": 175
     },
     {
-      "epoch": 0.15444015444015444,
-      "grad_norm": 4.358458042144775,
+      "epoch": 0.12622278321236982,
+      "grad_norm": 4.470991134643555,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.6081,
+      "loss": 0.6126,
       "step": 200
     },
     {
-      "epoch": 0.17374517374517376,
-      "grad_norm": 4.636687278747559,
+      "epoch": 0.14200063111391606,
+      "grad_norm": 4.39013147354126,
       "learning_rate": 4.5e-06,
-      "loss": 0.6034,
+      "loss": 0.5958,
       "step": 225
     },
     {
-      "epoch": 0.19305019305019305,
-      "grad_norm": 4.5222554206848145,
+      "epoch": 0.1577784790154623,
+      "grad_norm": 4.321168422698975,
       "learning_rate": 5e-06,
-      "loss": 0.5915,
+      "loss": 0.5697,
       "step": 250
     },
     {
-      "epoch": 0.21235521235521235,
-      "grad_norm": 4.840209007263184,
+      "epoch": 0.1735563269170085,
+      "grad_norm": 4.567933082580566,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.6055,
+      "loss": 0.5554,
       "step": 275
     },
     {
-      "epoch": 0.23166023166023167,
-      "grad_norm": 4.3733673095703125,
+      "epoch": 0.18933417481855475,
+      "grad_norm": 4.7674150466918945,
       "learning_rate": 6e-06,
-      "loss": 0.565,
+      "loss": 0.5646,
       "step": 300
     },
     {
-      "epoch": 0.25096525096525096,
-      "grad_norm": 3.8710694313049316,
+      "epoch": 0.20511202272010098,
+      "grad_norm": 4.203701496124268,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.5128,
+      "loss": 0.5316,
       "step": 325
     },
     {
-      "epoch": 0.2702702702702703,
-      "grad_norm": 4.533177375793457,
+      "epoch": 0.2208898706216472,
+      "grad_norm": 4.289330005645752,
       "learning_rate": 7e-06,
-      "loss": 0.5532,
+      "loss": 0.527,
       "step": 350
     },
     {
-      "epoch": 0.28957528957528955,
-      "grad_norm": 4.626265048980713,
+      "epoch": 0.23666771852319343,
+      "grad_norm": 4.419729232788086,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.5708,
+      "loss": 0.533,
       "step": 375
     },
     {
-      "epoch": 0.3088803088803089,
-      "grad_norm": 5.235887050628662,
+      "epoch": 0.25244556642473964,
+      "grad_norm": 3.873534917831421,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.5413,
+      "loss": 0.5164,
       "step": 400
     },
     {
-      "epoch": 0.3281853281853282,
-      "grad_norm": 4.232483386993408,
+      "epoch": 0.2682234143262859,
+      "grad_norm": 3.9782721996307373,
       "learning_rate": 8.5e-06,
-      "loss": 0.5326,
+      "loss": 0.4957,
       "step": 425
     },
     {
-      "epoch": 0.3474903474903475,
-      "grad_norm": 4.374448776245117,
+      "epoch": 0.2840012622278321,
+      "grad_norm": 4.02784538269043,
       "learning_rate": 9e-06,
-      "loss": 0.5205,
+      "loss": 0.5142,
       "step": 450
     },
     {
-      "epoch": 0.3667953667953668,
-      "grad_norm": 4.335531711578369,
+      "epoch": 0.29977911012937836,
+      "grad_norm": 4.09860372543335,
       "learning_rate": 9.5e-06,
-      "loss": 0.5347,
+      "loss": 0.4987,
       "step": 475
     },
     {
-      "epoch": 0.3861003861003861,
-      "grad_norm": 4.169550895690918,
+      "epoch": 0.3155569580309246,
+      "grad_norm": 3.7876222133636475,
       "learning_rate": 1e-05,
-      "loss": 0.5148,
+      "loss": 0.5076,
       "step": 500
     },
     {
-      "epoch": 0.40540540540540543,
-      "grad_norm": 3.5508618354797363,
+      "epoch": 0.33133480593247083,
+      "grad_norm": 4.0808916091918945,
       "learning_rate": 9.944444444444445e-06,
-      "loss": 0.5227,
+      "loss": 0.4824,
       "step": 525
     },
     {
-      "epoch": 0.4247104247104247,
-      "grad_norm": 3.7011241912841797,
+      "epoch": 0.347112653834017,
+      "grad_norm": 3.7431557178497314,
       "learning_rate": 9.88888888888889e-06,
-      "loss": 0.4989,
+      "loss": 0.4829,
       "step": 550
     },
     {
-      "epoch": 0.444015444015444,
-      "grad_norm": 3.37355899810791,
+      "epoch": 0.36289050173556325,
+      "grad_norm": 4.337482452392578,
       "learning_rate": 9.833333333333333e-06,
-      "loss": 0.5067,
+      "loss": 0.5014,
       "step": 575
     },
     {
-      "epoch": 0.46332046332046334,
-      "grad_norm": 4.071038246154785,
+      "epoch": 0.3786683496371095,
+      "grad_norm": 4.180824279785156,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 0.4884,
+      "loss": 0.4592,
       "step": 600
     },
     {
-      "epoch": 0.4826254826254826,
-      "grad_norm": 3.6045796871185303,
+      "epoch": 0.39444619753865573,
+      "grad_norm": 4.251559257507324,
       "learning_rate": 9.722222222222223e-06,
-      "loss": 0.4752,
+      "loss": 0.4665,
       "step": 625
     },
     {
-      "epoch": 0.5019305019305019,
-      "grad_norm": 3.414005994796753,
+      "epoch": 0.41022404544020197,
+      "grad_norm": 3.684410333633423,
       "learning_rate": 9.666666666666667e-06,
-      "loss": 0.4784,
+      "loss": 0.4652,
       "step": 650
     },
     {
-      "epoch": 0.5212355212355212,
-      "grad_norm": 3.203928232192993,
+      "epoch": 0.4260018933417482,
+      "grad_norm": 3.8727266788482666,
       "learning_rate": 9.611111111111112e-06,
-      "loss": 0.4915,
+      "loss": 0.44,
       "step": 675
     },
     {
-      "epoch": 0.5405405405405406,
-      "grad_norm": 4.148367404937744,
+      "epoch": 0.4417797412432944,
+      "grad_norm": 3.3472304344177246,
       "learning_rate": 9.555555555555556e-06,
-      "loss": 0.4861,
+      "loss": 0.4557,
       "step": 700
     },
     {
-      "epoch": 0.5598455598455598,
-      "grad_norm": 3.9569835662841797,
+      "epoch": 0.4575575891448406,
+      "grad_norm": 3.535940170288086,
       "learning_rate": 9.5e-06,
-      "loss": 0.4744,
+      "loss": 0.4449,
       "step": 725
     },
     {
-      "epoch": 0.5791505791505791,
-      "grad_norm": 3.682847261428833,
+      "epoch": 0.47333543704638686,
+      "grad_norm": 4.009454250335693,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 0.4662,
+      "loss": 0.431,
       "step": 750
     },
     {
-      "epoch": 0.5984555984555985,
-      "grad_norm": 3.498784303665161,
+      "epoch": 0.4891132849479331,
+      "grad_norm": 3.5649003982543945,
       "learning_rate": 9.38888888888889e-06,
-      "loss": 0.4479,
+      "loss": 0.4506,
       "step": 775
     },
     {
-      "epoch": 0.6177606177606177,
-      "grad_norm": 3.3415608406066895,
+      "epoch": 0.5048911328494793,
+      "grad_norm": 3.044846296310425,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 0.4422,
+      "loss": 0.427,
       "step": 800
     },
     {
-      "epoch": 0.637065637065637,
-      "grad_norm": 3.7531261444091797,
+      "epoch": 0.5206689807510255,
+      "grad_norm": 3.3232269287109375,
       "learning_rate": 9.277777777777778e-06,
-      "loss": 0.4639,
+      "loss": 0.4009,
       "step": 825
     },
     {
-      "epoch": 0.6563706563706564,
-      "grad_norm": 3.1370999813079834,
+      "epoch": 0.5364468286525718,
+      "grad_norm": 3.964977741241455,
       "learning_rate": 9.222222222222224e-06,
-      "loss": 0.4286,
+      "loss": 0.4319,
       "step": 850
     },
     {
-      "epoch": 0.6756756756756757,
-      "grad_norm": 3.474857807159424,
+      "epoch": 0.552224676554118,
+      "grad_norm": 3.00532865524292,
       "learning_rate": 9.166666666666666e-06,
-      "loss": 0.4461,
+      "loss": 0.4082,
       "step": 875
     },
     {
-      "epoch": 0.694980694980695,
-      "grad_norm": 3.5445330142974854,
+      "epoch": 0.5680025244556642,
+      "grad_norm": 3.6661252975463867,
       "learning_rate": 9.111111111111112e-06,
-      "loss": 0.4364,
+      "loss": 0.4171,
       "step": 900
     },
     {
-      "epoch": 0.7142857142857143,
-      "grad_norm": 3.309083938598633,
+      "epoch": 0.5837803723572105,
+      "grad_norm": 2.911687135696411,
       "learning_rate": 9.055555555555556e-06,
-      "loss": 0.4274,
+      "loss": 0.405,
       "step": 925
     },
     {
-      "epoch": 0.7335907335907336,
-      "grad_norm": 3.169285297393799,
+      "epoch": 0.5995582202587567,
+      "grad_norm": 3.7377912998199463,
       "learning_rate": 9e-06,
-      "loss": 0.4244,
+      "loss": 0.4032,
       "step": 950
     },
     {
-      "epoch": 0.752895752895753,
-      "grad_norm": 3.5987651348114014,
+      "epoch": 0.615336068160303,
+      "grad_norm": 3.389277696609497,
       "learning_rate": 8.944444444444446e-06,
-      "loss": 0.4189,
+      "loss": 0.3968,
       "step": 975
     },
     {
-      "epoch": 0.7722007722007722,
-      "grad_norm": 3.622044563293457,
+      "epoch": 0.6311139160618492,
+      "grad_norm": 3.2546145915985107,
       "learning_rate": 8.888888888888888e-06,
-      "loss": 0.4118,
+      "loss": 0.408,
       "step": 1000
     },
     {
-      "epoch": 0.7722007722007722,
-      "eval_loss": 0.48980528116226196,
-      "eval_runtime": 2996.853,
-      "eval_samples_per_second": 2.343,
+      "epoch": 0.6311139160618492,
+      "eval_loss": 0.5,
+      "eval_runtime": 3003.1872,
+      "eval_samples_per_second": 2.338,
       "eval_steps_per_second": 0.146,
-      "eval_wer": 0.3626803079679568,
+      "eval_wer": 0.3734339502318352,
       "step": 1000
     },
     {
-      "epoch": 0.7915057915057915,
-      "grad_norm": 3.3155651092529297,
+      "epoch": 0.6468917639633954,
+      "grad_norm": 3.6185717582702637,
       "learning_rate": 8.833333333333334e-06,
-      "loss": 0.4394,
+      "loss": 0.4092,
       "step": 1025
     },
     {
-      "epoch": 0.8108108108108109,
-      "grad_norm": 2.6345057487487793,
+      "epoch": 0.6626696118649417,
+      "grad_norm": 3.227266788482666,
       "learning_rate": 8.777777777777778e-06,
-      "loss": 0.4143,
+      "loss": 0.3839,
       "step": 1050
     },
     {
-      "epoch": 0.8301158301158301,
-      "grad_norm": 3.3825740814208984,
+      "epoch": 0.6784474597664879,
+      "grad_norm": 3.281687021255493,
       "learning_rate": 8.722222222222224e-06,
-      "loss": 0.399,
+      "loss": 0.3954,
       "step": 1075
     },
     {
-      "epoch": 0.8494208494208494,
-      "grad_norm": 3.439831256866455,
+      "epoch": 0.694225307668034,
+      "grad_norm": 3.359879732131958,
       "learning_rate": 8.666666666666668e-06,
-      "loss": 0.4269,
+      "loss": 0.4051,
       "step": 1100
     },
     {
-      "epoch": 0.8687258687258688,
-      "grad_norm": 2.9175798892974854,
+      "epoch": 0.7100031555695803,
+      "grad_norm": 2.7654247283935547,
       "learning_rate": 8.611111111111112e-06,
-      "loss": 0.46,
+      "loss": 0.3979,
       "step": 1125
     },
     {
-      "epoch": 0.888030888030888,
-      "grad_norm": 3.6642494201660156,
+      "epoch": 0.7257810034711265,
+      "grad_norm": 2.9193856716156006,
       "learning_rate": 8.555555555555556e-06,
-      "loss": 0.4038,
+      "loss": 0.4075,
       "step": 1150
     },
     {
-      "epoch": 0.9073359073359073,
-      "grad_norm": 3.3490536212921143,
+      "epoch": 0.7415588513726727,
+      "grad_norm": 3.3058297634124756,
       "learning_rate": 8.5e-06,
-      "loss": 0.3989,
+      "loss": 0.4176,
       "step": 1175
     },
     {
-      "epoch": 0.9266409266409267,
-      "grad_norm": 3.563148021697998,
+      "epoch": 0.757336699274219,
+      "grad_norm": 3.126009941101074,
       "learning_rate": 8.444444444444446e-06,
-      "loss": 0.4221,
+      "loss": 0.3928,
       "step": 1200
     },
     {
-      "epoch": 0.9459459459459459,
-      "grad_norm": 3.2811503410339355,
+      "epoch": 0.7731145471757652,
+      "grad_norm": 2.937455415725708,
       "learning_rate": 8.38888888888889e-06,
-      "loss": 0.3959,
+      "loss": 0.3772,
       "step": 1225
     },
     {
-      "epoch": 0.9652509652509652,
-      "grad_norm": 4.630711078643799,
+      "epoch": 0.7888923950773115,
+      "grad_norm": 2.8536481857299805,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.3992,
+      "loss": 0.3968,
       "step": 1250
     },
     {
-      "epoch": 0.9845559845559846,
-      "grad_norm": 2.9398629665374756,
+      "epoch": 0.8046702429788577,
+      "grad_norm": 3.651503086090088,
       "learning_rate": 8.277777777777778e-06,
-      "loss": 0.4075,
+      "loss": 0.3758,
       "step": 1275
     },
     {
-      "epoch": 1.0038610038610039,
-      "grad_norm": 2.4827306270599365,
+      "epoch": 0.8204480908804039,
+      "grad_norm": 2.8864688873291016,
       "learning_rate": 8.222222222222222e-06,
-      "loss": 0.3593,
+      "loss": 0.3784,
       "step": 1300
     },
     {
-      "epoch": 1.0231660231660231,
-      "grad_norm": 2.922198534011841,
+      "epoch": 0.8362259387819502,
+      "grad_norm": 3.2489211559295654,
       "learning_rate": 8.166666666666668e-06,
-      "loss": 0.2864,
+      "loss": 0.3632,
       "step": 1325
     },
     {
-      "epoch": 1.0424710424710424,
-      "grad_norm": 2.595409393310547,
+      "epoch": 0.8520037866834964,
+      "grad_norm": 3.0353879928588867,
       "learning_rate": 8.111111111111112e-06,
-      "loss": 0.2928,
+      "loss": 0.3564,
       "step": 1350
     },
     {
-      "epoch": 1.0617760617760619,
-      "grad_norm": 2.6854002475738525,
+      "epoch": 0.8677816345850426,
+      "grad_norm": 3.2296388149261475,
       "learning_rate": 8.055555555555557e-06,
-      "loss": 0.2773,
+      "loss": 0.3546,
       "step": 1375
     },
     {
-      "epoch": 1.0810810810810811,
-      "grad_norm": 2.5066304206848145,
+      "epoch": 0.8835594824865888,
+      "grad_norm": 3.0037214756011963,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.2864,
+      "loss": 0.3453,
       "step": 1400
     },
     {
-      "epoch": 1.1003861003861004,
-      "grad_norm": 2.477524518966675,
+      "epoch": 0.899337330388135,
+      "grad_norm": 3.308013439178467,
       "learning_rate": 7.944444444444445e-06,
-      "loss": 0.2815,
+      "loss": 0.3578,
       "step": 1425
     },
     {
-      "epoch": 1.1196911196911197,
-      "grad_norm": 2.947479724884033,
+      "epoch": 0.9151151782896813,
+      "grad_norm": 3.3210103511810303,
       "learning_rate": 7.88888888888889e-06,
-      "loss": 0.2814,
+      "loss": 0.3382,
       "step": 1450
     },
     {
-      "epoch": 1.138996138996139,
-      "grad_norm": 2.497398853302002,
+      "epoch": 0.9308930261912275,
+      "grad_norm": 3.2565624713897705,
       "learning_rate": 7.833333333333333e-06,
-      "loss": 0.27,
+      "loss": 0.3718,
       "step": 1475
     },
     {
-      "epoch": 1.1583011583011582,
-      "grad_norm": 2.5605263710021973,
+      "epoch": 0.9466708740927737,
+      "grad_norm": 3.0643301010131836,
       "learning_rate": 7.77777777777778e-06,
-      "loss": 0.2778,
+      "loss": 0.3478,
       "step": 1500
     },
     {
-      "epoch": 1.1776061776061777,
-      "grad_norm": 2.8643362522125244,
+      "epoch": 0.96244872199432,
+      "grad_norm": 2.628441333770752,
       "learning_rate": 7.722222222222223e-06,
-      "loss": 0.2625,
+      "loss": 0.3654,
       "step": 1525
     },
     {
-      "epoch": 1.196911196911197,
-      "grad_norm": 2.8837692737579346,
+      "epoch": 0.9782265698958662,
+      "grad_norm": 2.842191457748413,
       "learning_rate": 7.666666666666667e-06,
-      "loss": 0.2732,
+      "loss": 0.3727,
       "step": 1550
     },
     {
-      "epoch": 1.2162162162162162,
-      "grad_norm": 2.499385118484497,
+      "epoch": 0.9940044177974124,
+      "grad_norm": 2.23713755607605,
       "learning_rate": 7.611111111111111e-06,
-      "loss": 0.2898,
+      "loss": 0.3369,
       "step": 1575
     },
     {
-      "epoch": 1.2355212355212355,
-      "grad_norm": 3.035726547241211,
+      "epoch": 1.0097822656989586,
+      "grad_norm": 2.4432640075683594,
       "learning_rate": 7.555555555555556e-06,
-      "loss": 0.279,
+      "loss": 0.2744,
       "step": 1600
     },
     {
-      "epoch": 1.2548262548262548,
-      "grad_norm": 2.872089385986328,
+      "epoch": 1.0255601136005048,
+      "grad_norm": 2.3971242904663086,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.3005,
+      "loss": 0.2449,
       "step": 1625
     },
     {
-      "epoch": 1.2741312741312742,
-      "grad_norm": 2.8907766342163086,
+      "epoch": 1.041337961502051,
+      "grad_norm": 2.4431796073913574,
       "learning_rate": 7.444444444444445e-06,
-      "loss": 0.2941,
+      "loss": 0.2499,
       "step": 1650
     },
     {
-      "epoch": 1.2934362934362935,
-      "grad_norm": 2.8525350093841553,
+      "epoch": 1.0571158094035973,
+      "grad_norm": 2.3628671169281006,
       "learning_rate": 7.38888888888889e-06,
-      "loss": 0.2644,
+      "loss": 0.2698,
       "step": 1675
     },
     {
-      "epoch": 1.3127413127413128,
-      "grad_norm": 2.7391347885131836,
+      "epoch": 1.0728936573051435,
+      "grad_norm": 2.0304720401763916,
       "learning_rate": 7.333333333333333e-06,
-      "loss": 0.2811,
+      "loss": 0.2418,
       "step": 1700
     },
     {
-      "epoch": 1.332046332046332,
-      "grad_norm": 2.4943618774414062,
+      "epoch": 1.0886715052066898,
+      "grad_norm": 2.804999828338623,
       "learning_rate": 7.277777777777778e-06,
-      "loss": 0.2786,
+      "loss": 0.2506,
       "step": 1725
     },
     {
-      "epoch": 1.3513513513513513,
-      "grad_norm": 3.009016752243042,
+      "epoch": 1.104449353108236,
+      "grad_norm": 3.027466297149658,
       "learning_rate": 7.222222222222223e-06,
-      "loss": 0.2928,
+      "loss": 0.2568,
       "step": 1750
     },
     {
-      "epoch": 1.3706563706563706,
-      "grad_norm": 3.0978188514709473,
+      "epoch": 1.1202272010097822,
+      "grad_norm": 2.6216728687286377,
       "learning_rate": 7.166666666666667e-06,
-      "loss": 0.298,
+      "loss": 0.2417,
       "step": 1775
     },
     {
-      "epoch": 1.3899613899613898,
-      "grad_norm": 2.746687412261963,
+      "epoch": 1.1360050489113285,
+      "grad_norm": 2.6749463081359863,
       "learning_rate": 7.111111111111112e-06,
-      "loss": 0.2848,
+      "loss": 0.2479,
       "step": 1800
     },
     {
-      "epoch": 1.4092664092664093,
-      "grad_norm": 2.6545677185058594,
+      "epoch": 1.1517828968128747,
+      "grad_norm": 2.4260010719299316,
       "learning_rate": 7.055555555555557e-06,
-      "loss": 0.2755,
+      "loss": 0.2385,
       "step": 1825
     },
     {
-      "epoch": 1.4285714285714286,
-      "grad_norm": 3.196070671081543,
+      "epoch": 1.167560744714421,
+      "grad_norm": 2.3928089141845703,
       "learning_rate": 7e-06,
-      "loss": 0.2733,
+      "loss": 0.2321,
       "step": 1850
     },
     {
-      "epoch": 1.4478764478764479,
-      "grad_norm": 2.5609796047210693,
+      "epoch": 1.1833385926159672,
+      "grad_norm": 2.639481782913208,
       "learning_rate": 6.944444444444445e-06,
-      "loss": 0.2659,
+      "loss": 0.2429,
       "step": 1875
     },
     {
-      "epoch": 1.4671814671814671,
-      "grad_norm": 2.339729070663452,
+      "epoch": 1.1991164405175134,
+      "grad_norm": 3.0133748054504395,
       "learning_rate": 6.88888888888889e-06,
-      "loss": 0.2924,
+      "loss": 0.252,
       "step": 1900
     },
     {
-      "epoch": 1.4864864864864864,
-      "grad_norm": 2.7920401096343994,
+      "epoch": 1.2148942884190597,
+      "grad_norm": 2.657839059829712,
       "learning_rate": 6.833333333333334e-06,
-      "loss": 0.2628,
+      "loss": 0.2453,
       "step": 1925
     },
     {
-      "epoch": 1.505791505791506,
-      "grad_norm": 2.2347285747528076,
+      "epoch": 1.230672136320606,
+      "grad_norm": 2.745032787322998,
       "learning_rate": 6.777777777777779e-06,
-      "loss": 0.2816,
+      "loss": 0.2454,
       "step": 1950
     },
     {
-      "epoch": 1.525096525096525,
-      "grad_norm": 2.6325347423553467,
+      "epoch": 1.2464499842221521,
+      "grad_norm": 2.7123420238494873,
       "learning_rate": 6.7222222222222235e-06,
-      "loss": 0.2664,
+      "loss": 0.2672,
       "step": 1975
     },
     {
-      "epoch": 1.5444015444015444,
-      "grad_norm": 3.093266248703003,
+      "epoch": 1.2622278321236984,
+      "grad_norm": 2.789599657058716,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.2629,
+      "loss": 0.2455,
       "step": 2000
     },
     {
-      "epoch": 1.5444015444015444,
-      "eval_loss": 0.4272039532661438,
-      "eval_runtime": 2939.2445,
-      "eval_samples_per_second": 2.389,
-      "eval_steps_per_second": 0.149,
-      "eval_wer": 0.3144726762830916,
+      "epoch": 1.2622278321236984,
+      "eval_loss": 0.4339774549007416,
+      "eval_runtime": 2958.2023,
+      "eval_samples_per_second": 2.374,
+      "eval_steps_per_second": 0.148,
+      "eval_wer": 0.32330707196853953,
       "step": 2000
     },
     {
-      "epoch": 1.5637065637065637,
-      "grad_norm": 2.7231810092926025,
+      "epoch": 1.2780056800252446,
+      "grad_norm": 3.1237881183624268,
       "learning_rate": 6.6111111111111115e-06,
-      "loss": 0.2671,
+      "loss": 0.2584,
       "step": 2025
     },
     {
-      "epoch": 1.583011583011583,
-      "grad_norm": 2.4896557331085205,
+      "epoch": 1.2937835279267909,
+      "grad_norm": 2.1236650943756104,
       "learning_rate": 6.555555555555556e-06,
-      "loss": 0.291,
+      "loss": 0.2289,
       "step": 2050
     },
     {
-      "epoch": 1.6023166023166024,
-      "grad_norm": 2.549482583999634,
+      "epoch": 1.309561375828337,
+      "grad_norm": 3.098536968231201,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.2697,
+      "loss": 0.2392,
       "step": 2075
     },
     {
-      "epoch": 1.6216216216216215,
-      "grad_norm": 3.055518627166748,
+      "epoch": 1.325339223729883,
+      "grad_norm": 2.5509822368621826,
       "learning_rate": 6.444444444444445e-06,
-      "loss": 0.2612,
+      "loss": 0.2298,
       "step": 2100
     },
     {
-      "epoch": 1.640926640926641,
-      "grad_norm": 2.8023324012756348,
+      "epoch": 1.3411170716314293,
+      "grad_norm": 2.699902296066284,
       "learning_rate": 6.3888888888888885e-06,
-      "loss": 0.2612,
+      "loss": 0.2425,
       "step": 2125
     },
     {
-      "epoch": 1.6602316602316602,
-      "grad_norm": 2.5764496326446533,
+      "epoch": 1.3568949195329756,
+      "grad_norm": 2.5050745010375977,
       "learning_rate": 6.333333333333333e-06,
-      "loss": 0.2718,
+      "loss": 0.2377,
       "step": 2150
     },
     {
-      "epoch": 1.6795366795366795,
-      "grad_norm": 3.0034117698669434,
+      "epoch": 1.3726727674345218,
+      "grad_norm": 2.2431986331939697,
       "learning_rate": 6.277777777777778e-06,
-      "loss": 0.2836,
+      "loss": 0.225,
       "step": 2175
     },
     {
-      "epoch": 1.698841698841699,
-      "grad_norm": 3.285400629043579,
+      "epoch": 1.388450615336068,
+      "grad_norm": 2.3476831912994385,
       "learning_rate": 6.222222222222223e-06,
-      "loss": 0.267,
+      "loss": 0.2458,
       "step": 2200
     },
     {
-      "epoch": 1.718146718146718,
-      "grad_norm": 2.8904147148132324,
+      "epoch": 1.4042284632376143,
+      "grad_norm": 2.1803078651428223,
       "learning_rate": 6.166666666666667e-06,
-      "loss": 0.2844,
+      "loss": 0.2459,
       "step": 2225
     },
     {
-      "epoch": 1.7374517374517375,
-      "grad_norm": 2.7520534992218018,
+      "epoch": 1.4200063111391605,
+      "grad_norm": 2.475566864013672,
       "learning_rate": 6.111111111111112e-06,
-      "loss": 0.2641,
+      "loss": 0.2349,
       "step": 2250
     },
     {
-      "epoch": 1.7567567567567568,
-      "grad_norm": 2.262890338897705,
+      "epoch": 1.4357841590407068,
+      "grad_norm": 2.735623598098755,
       "learning_rate": 6.055555555555555e-06,
-      "loss": 0.2636,
+      "loss": 0.2295,
       "step": 2275
     },
     {
-      "epoch": 1.776061776061776,
-      "grad_norm": 2.410085439682007,
+      "epoch": 1.451562006942253,
+      "grad_norm": 2.5407278537750244,
       "learning_rate": 6e-06,
-      "loss": 0.2753,
+      "loss": 0.2367,
       "step": 2300
     },
     {
-      "epoch": 1.7953667953667953,
-      "grad_norm": 2.5720226764678955,
+      "epoch": 1.4673398548437993,
+      "grad_norm": 2.189357042312622,
       "learning_rate": 5.944444444444445e-06,
-      "loss": 0.2702,
+      "loss": 0.2367,
       "step": 2325
     },
     {
-      "epoch": 1.8146718146718146,
-      "grad_norm": 2.999687433242798,
+      "epoch": 1.4831177027453455,
+      "grad_norm": 2.2761070728302,
       "learning_rate": 5.88888888888889e-06,
-      "loss": 0.2998,
+      "loss": 0.2331,
       "step": 2350
     },
     {
-      "epoch": 1.833976833976834,
-      "grad_norm": 2.5406405925750732,
+      "epoch": 1.4988955506468917,
+      "grad_norm": 2.657602071762085,
       "learning_rate": 5.833333333333334e-06,
-      "loss": 0.2444,
+      "loss": 0.2346,
       "step": 2375
     },
     {
-      "epoch": 1.8532818532818531,
-      "grad_norm": 2.314110517501831,
+      "epoch": 1.514673398548438,
+      "grad_norm": 2.7713756561279297,
       "learning_rate": 5.777777777777778e-06,
-      "loss": 0.2664,
+      "loss": 0.2336,
       "step": 2400
     },
     {
-      "epoch": 1.8725868725868726,
-      "grad_norm": 2.745534896850586,
+      "epoch": 1.5304512464499842,
+      "grad_norm": 2.4739747047424316,
       "learning_rate": 5.722222222222222e-06,
-      "loss": 0.2676,
+      "loss": 0.2314,
       "step": 2425
     },
     {
-      "epoch": 1.8918918918918919,
-      "grad_norm": 3.1628456115722656,
+      "epoch": 1.5462290943515304,
+      "grad_norm": 2.275024175643921,
       "learning_rate": 5.666666666666667e-06,
-      "loss": 0.2475,
+      "loss": 0.2384,
       "step": 2450
     },
     {
-      "epoch": 1.9111969111969112,
-      "grad_norm": 2.6415674686431885,
+      "epoch": 1.5620069422530767,
+      "grad_norm": 8.249900817871094,
       "learning_rate": 5.611111111111112e-06,
-      "loss": 0.2565,
+      "loss": 0.2137,
       "step": 2475
     },
     {
-      "epoch": 1.9305019305019306,
-      "grad_norm": 2.4771909713745117,
+      "epoch": 1.577784790154623,
+      "grad_norm": 2.453125,
       "learning_rate": 5.555555555555557e-06,
-      "loss": 0.2621,
+      "loss": 0.2445,
       "step": 2500
     },
     {
-      "epoch": 1.9498069498069497,
-      "grad_norm": 2.7251639366149902,
+      "epoch": 1.5935626380561692,
+      "grad_norm": 2.3670058250427246,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.2703,
+      "loss": 0.2418,
       "step": 2525
     },
     {
-      "epoch": 1.9691119691119692,
-      "grad_norm": 2.73382830619812,
+      "epoch": 1.6093404859577154,
+      "grad_norm": 2.700669527053833,
       "learning_rate": 5.444444444444445e-06,
-      "loss": 0.2728,
+      "loss": 0.2356,
       "step": 2550
     },
     {
-      "epoch": 1.9884169884169884,
-      "grad_norm": 3.0017306804656982,
+      "epoch": 1.6251183338592616,
+      "grad_norm": 2.0962114334106445,
       "learning_rate": 5.388888888888889e-06,
-      "loss": 0.2521,
+      "loss": 0.2268,
       "step": 2575
     },
     {
-      "epoch": 2.0077220077220077,
-      "grad_norm": 1.9712167978286743,
+      "epoch": 1.6408961817608079,
+      "grad_norm": 2.2385094165802,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 0.2352,
+      "loss": 0.2292,
       "step": 2600
     },
     {
-      "epoch": 2.027027027027027,
-      "grad_norm": 1.8378448486328125,
+      "epoch": 1.656674029662354,
+      "grad_norm": 2.2060489654541016,
       "learning_rate": 5.2777777777777785e-06,
-      "loss": 0.1696,
+      "loss": 0.2255,
       "step": 2625
     },
     {
-      "epoch": 2.0463320463320462,
-      "grad_norm": 2.419578790664673,
+      "epoch": 1.6724518775639003,
+      "grad_norm": 2.859130859375,
       "learning_rate": 5.2222222222222226e-06,
-      "loss": 0.1592,
+      "loss": 0.2304,
       "step": 2650
     },
     {
-      "epoch": 2.0656370656370657,
-      "grad_norm": 2.145256757736206,
+      "epoch": 1.6882297254654466,
+      "grad_norm": 2.400465250015259,
       "learning_rate": 5.1666666666666675e-06,
-      "loss": 0.1719,
+      "loss": 0.2375,
       "step": 2675
     },
     {
-      "epoch": 2.0849420849420848,
-      "grad_norm": 2.147104024887085,
+      "epoch": 1.7040075733669928,
+      "grad_norm": 2.382413387298584,
       "learning_rate": 5.1111111111111115e-06,
-      "loss": 0.1487,
+      "loss": 0.2212,
       "step": 2700
     },
     {
-      "epoch": 2.1042471042471043,
-      "grad_norm": 2.280949592590332,
+      "epoch": 1.719785421268539,
+      "grad_norm": 2.3223814964294434,
       "learning_rate": 5.0555555555555555e-06,
-      "loss": 0.1785,
+      "loss": 0.2274,
       "step": 2725
     },
     {
-      "epoch": 2.1235521235521237,
-      "grad_norm": 2.359393835067749,
+      "epoch": 1.7355632691700853,
+      "grad_norm": 2.5504796504974365,
       "learning_rate": 5e-06,
-      "loss": 0.1656,
+      "loss": 0.2224,
       "step": 2750
     },
     {
-      "epoch": 2.142857142857143,
-      "grad_norm": 2.4849932193756104,
+      "epoch": 1.7513411170716315,
+      "grad_norm": 1.8646706342697144,
       "learning_rate": 4.944444444444445e-06,
-      "loss": 0.1697,
+      "loss": 0.2109,
       "step": 2775
     },
     {
-      "epoch": 2.1621621621621623,
-      "grad_norm": 2.478525400161743,
+      "epoch": 1.7671189649731778,
+      "grad_norm": 2.2399260997772217,
       "learning_rate": 4.888888888888889e-06,
-      "loss": 0.1586,
+      "loss": 0.2253,
       "step": 2800
     },
     {
-      "epoch": 2.1814671814671813,
-      "grad_norm": 2.258152723312378,
+      "epoch": 1.782896812874724,
+      "grad_norm": 2.5138754844665527,
       "learning_rate": 4.833333333333333e-06,
-      "loss": 0.1677,
+      "loss": 0.2226,
       "step": 2825
     },
     {
-      "epoch": 2.200772200772201,
-      "grad_norm": 2.2826361656188965,
+      "epoch": 1.7986746607762703,
+      "grad_norm": 2.8491804599761963,
       "learning_rate": 4.777777777777778e-06,
-      "loss": 0.1603,
+      "loss": 0.2281,
       "step": 2850
     },
     {
-      "epoch": 2.2200772200772203,
-      "grad_norm": 2.3771138191223145,
+      "epoch": 1.8144525086778165,
+      "grad_norm": 2.52425217628479,
       "learning_rate": 4.722222222222222e-06,
-      "loss": 0.159,
+      "loss": 0.222,
       "step": 2875
     },
     {
-      "epoch": 2.2393822393822393,
-      "grad_norm": 2.353187322616577,
+      "epoch": 1.8302303565793627,
+      "grad_norm": 2.4254088401794434,
       "learning_rate": 4.666666666666667e-06,
-      "loss": 0.1617,
+      "loss": 0.216,
       "step": 2900
     },
     {
-      "epoch": 2.258687258687259,
-      "grad_norm": 2.2030959129333496,
+      "epoch": 1.846008204480909,
+      "grad_norm": 2.4229941368103027,
       "learning_rate": 4.611111111111112e-06,
-      "loss": 0.17,
+      "loss": 0.2178,
       "step": 2925
     },
     {
-      "epoch": 2.277992277992278,
-      "grad_norm": 2.1587460041046143,
+      "epoch": 1.861786052382455,
+      "grad_norm": 2.6721200942993164,
       "learning_rate": 4.555555555555556e-06,
-      "loss": 0.1618,
+      "loss": 0.2241,
       "step": 2950
     },
     {
-      "epoch": 2.2972972972972974,
-      "grad_norm": 2.918156623840332,
+      "epoch": 1.8775639002840012,
+      "grad_norm": 2.3381521701812744,
       "learning_rate": 4.5e-06,
-      "loss": 0.1545,
+      "loss": 0.2255,
       "step": 2975
     },
     {
-      "epoch": 2.3166023166023164,
-      "grad_norm": 2.4041364192962646,
+      "epoch": 1.8933417481855475,
+      "grad_norm": 2.385972023010254,
       "learning_rate": 4.444444444444444e-06,
-      "loss": 0.1677,
+      "loss": 0.216,
       "step": 3000
     },
     {
-      "epoch": 2.3166023166023164,
-      "eval_loss": 0.42425960302352905,
-      "eval_runtime": 2930.947,
-      "eval_samples_per_second": 2.396,
-      "eval_steps_per_second": 0.15,
-      "eval_wer": 0.3038651796263357,
+      "epoch": 1.8933417481855475,
+      "eval_loss": 0.40097150206565857,
+      "eval_runtime": 2940.5315,
+      "eval_samples_per_second": 2.388,
+      "eval_steps_per_second": 0.149,
+      "eval_wer": 0.3000305570686471,
       "step": 3000
     },
     {
-      "epoch": 2.335907335907336,
-      "grad_norm": 1.8374619483947754,
+      "epoch": 1.9091195960870937,
+      "grad_norm": 2.507140636444092,
       "learning_rate": 4.388888888888889e-06,
-      "loss": 0.1505,
+      "loss": 0.2279,
       "step": 3025
     },
     {
-      "epoch": 2.3552123552123554,
-      "grad_norm": 2.2818222045898438,
+      "epoch": 1.92489744398864,
+      "grad_norm": 2.303067922592163,
       "learning_rate": 4.333333333333334e-06,
-      "loss": 0.1635,
+      "loss": 0.2158,
       "step": 3050
     },
     {
-      "epoch": 2.3745173745173744,
-      "grad_norm": 2.3476946353912354,
+      "epoch": 1.9406752918901862,
+      "grad_norm": 2.6614551544189453,
       "learning_rate": 4.277777777777778e-06,
-      "loss": 0.1572,
+      "loss": 0.21,
       "step": 3075
     },
     {
-      "epoch": 2.393822393822394,
-      "grad_norm": 2.727679491043091,
+      "epoch": 1.9564531397917324,
+      "grad_norm": 2.401498794555664,
       "learning_rate": 4.222222222222223e-06,
-      "loss": 0.1567,
+      "loss": 0.2301,
       "step": 3100
     },
     {
-      "epoch": 2.413127413127413,
-      "grad_norm": 2.148946523666382,
+      "epoch": 1.9722309876932786,
+      "grad_norm": 2.147493839263916,
       "learning_rate": 4.166666666666667e-06,
-      "loss": 0.1629,
+      "loss": 0.2281,
       "step": 3125
     },
     {
-      "epoch": 2.4324324324324325,
-      "grad_norm": 2.3598289489746094,
+      "epoch": 1.9880088355948249,
+      "grad_norm": 2.534454584121704,
       "learning_rate": 4.111111111111111e-06,
-      "loss": 0.1537,
+      "loss": 0.2176,
       "step": 3150
     },
     {
-      "epoch": 2.4517374517374515,
-      "grad_norm": 2.627530813217163,
+      "epoch": 2.003786683496371,
+      "grad_norm": 1.805999994277954,
       "learning_rate": 4.055555555555556e-06,
-      "loss": 0.1617,
+      "loss": 0.2052,
       "step": 3175
     },
     {
-      "epoch": 2.471042471042471,
-      "grad_norm": 2.1012682914733887,
+      "epoch": 2.019564531397917,
+      "grad_norm": 2.5037739276885986,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.1541,
+      "loss": 0.1369,
       "step": 3200
     },
     {
-      "epoch": 2.4903474903474905,
-      "grad_norm": 2.719054937362671,
+      "epoch": 2.0353423792994634,
+      "grad_norm": 2.1041407585144043,
       "learning_rate": 3.944444444444445e-06,
-      "loss": 0.1479,
+      "loss": 0.1223,
       "step": 3225
     },
     {
-      "epoch": 2.5096525096525095,
-      "grad_norm": 1.9170578718185425,
+      "epoch": 2.0511202272010096,
+      "grad_norm": 1.7618393898010254,
       "learning_rate": 3.88888888888889e-06,
-      "loss": 0.1558,
+      "loss": 0.1321,
       "step": 3250
     },
     {
-      "epoch": 2.528957528957529,
-      "grad_norm": 2.209770917892456,
+      "epoch": 2.066898075102556,
+      "grad_norm": 2.2356462478637695,
       "learning_rate": 3.833333333333334e-06,
-      "loss": 0.1494,
+      "loss": 0.1309,
       "step": 3275
     },
     {
-      "epoch": 2.5482625482625485,
-      "grad_norm": 2.3125557899475098,
+      "epoch": 2.082675923004102,
+      "grad_norm": 1.9169081449508667,
       "learning_rate": 3.777777777777778e-06,
-      "loss": 0.1549,
+      "loss": 0.1248,
       "step": 3300
     },
     {
-      "epoch": 2.5675675675675675,
-      "grad_norm": 2.2932839393615723,
+      "epoch": 2.0984537709056483,
+      "grad_norm": 2.0058977603912354,
       "learning_rate": 3.7222222222222225e-06,
-      "loss": 0.1529,
+      "loss": 0.1295,
       "step": 3325
     },
     {
-      "epoch": 2.586872586872587,
-      "grad_norm": 2.347710371017456,
+      "epoch": 2.1142316188071946,
+      "grad_norm": 1.9169594049453735,
       "learning_rate": 3.6666666666666666e-06,
-      "loss": 0.1637,
+      "loss": 0.1307,
       "step": 3350
     },
     {
-      "epoch": 2.606177606177606,
-      "grad_norm": 2.479405641555786,
+      "epoch": 2.130009466708741,
+      "grad_norm": 1.9053386449813843,
       "learning_rate": 3.6111111111111115e-06,
-      "loss": 0.162,
+      "loss": 0.1171,
       "step": 3375
     },
     {
-      "epoch": 2.6254826254826256,
-      "grad_norm": 2.7078042030334473,
+      "epoch": 2.145787314610287,
+      "grad_norm": 2.123209238052368,
       "learning_rate": 3.555555555555556e-06,
-      "loss": 0.1606,
+      "loss": 0.1501,
       "step": 3400
     },
     {
-      "epoch": 2.6447876447876446,
-      "grad_norm": 2.097592830657959,
+      "epoch": 2.1615651625118333,
+      "grad_norm": 2.103135347366333,
       "learning_rate": 3.5e-06,
-      "loss": 0.1554,
+      "loss": 0.1286,
       "step": 3425
     },
     {
-      "epoch": 2.664092664092664,
-      "grad_norm": 2.405099630355835,
+      "epoch": 2.1773430104133795,
+      "grad_norm": 1.901191234588623,
       "learning_rate": 3.444444444444445e-06,
-      "loss": 0.1611,
+      "loss": 0.1328,
       "step": 3450
     },
     {
-      "epoch": 2.6833976833976836,
-      "grad_norm": 2.120976686477661,
+      "epoch": 2.1931208583149258,
+      "grad_norm": 1.752928376197815,
       "learning_rate": 3.3888888888888893e-06,
-      "loss": 0.1537,
+      "loss": 0.1476,
       "step": 3475
     },
     {
-      "epoch": 2.7027027027027026,
-      "grad_norm": 2.165222406387329,
+      "epoch": 2.208898706216472,
+      "grad_norm": 2.233229160308838,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.156,
+      "loss": 0.133,
       "step": 3500
     },
     {
-      "epoch": 2.722007722007722,
-      "grad_norm": 2.3541574478149414,
+      "epoch": 2.2246765541180182,
+      "grad_norm": 1.8100041151046753,
       "learning_rate": 3.277777777777778e-06,
-      "loss": 0.1571,
+      "loss": 0.1401,
       "step": 3525
     },
     {
-      "epoch": 2.741312741312741,
-      "grad_norm": 2.5069708824157715,
+      "epoch": 2.2404544020195645,
+      "grad_norm": 2.2920119762420654,
       "learning_rate": 3.2222222222222227e-06,
-      "loss": 0.1467,
+      "loss": 0.1339,
       "step": 3550
     },
     {
-      "epoch": 2.7606177606177607,
-      "grad_norm": 1.9154902696609497,
+      "epoch": 2.2562322499211107,
+      "grad_norm": 2.502577543258667,
       "learning_rate": 3.1666666666666667e-06,
-      "loss": 0.1556,
+      "loss": 0.1431,
       "step": 3575
     },
     {
-      "epoch": 2.7799227799227797,
-      "grad_norm": 1.964211106300354,
+      "epoch": 2.272010097822657,
+      "grad_norm": 2.1777825355529785,
       "learning_rate": 3.1111111111111116e-06,
-      "loss": 0.1534,
+      "loss": 0.1287,
       "step": 3600
     },
     {
-      "epoch": 2.799227799227799,
-      "grad_norm": 2.3793370723724365,
+      "epoch": 2.287787945724203,
+      "grad_norm": 2.158822536468506,
       "learning_rate": 3.055555555555556e-06,
-      "loss": 0.1578,
+      "loss": 0.1427,
       "step": 3625
     },
     {
-      "epoch": 2.8185328185328187,
-      "grad_norm": 2.7694272994995117,
+      "epoch": 2.3035657936257494,
+      "grad_norm": 2.028738260269165,
       "learning_rate": 3e-06,
-      "loss": 0.1518,
+      "loss": 0.1307,
       "step": 3650
     },
     {
-      "epoch": 2.8378378378378377,
-      "grad_norm": 2.1063687801361084,
+      "epoch": 2.3193436415272957,
+      "grad_norm": 2.283491849899292,
       "learning_rate": 2.944444444444445e-06,
-      "loss": 0.1534,
+      "loss": 0.1284,
       "step": 3675
     },
     {
-      "epoch": 2.857142857142857,
-      "grad_norm": 2.708767890930176,
+      "epoch": 2.335121489428842,
+      "grad_norm": 2.0819170475006104,
       "learning_rate": 2.888888888888889e-06,
-      "loss": 0.153,
+      "loss": 0.1341,
       "step": 3700
     },
     {
-      "epoch": 2.8764478764478767,
-      "grad_norm": 1.7426552772521973,
+      "epoch": 2.350899337330388,
+      "grad_norm": 2.2217302322387695,
       "learning_rate": 2.8333333333333335e-06,
-      "loss": 0.1501,
+      "loss": 0.1398,
       "step": 3725
     },
     {
-      "epoch": 2.8957528957528957,
-      "grad_norm": 2.721165180206299,
+      "epoch": 2.3666771852319344,
+      "grad_norm": 1.7553716897964478,
       "learning_rate": 2.7777777777777783e-06,
-      "loss": 0.1613,
+      "loss": 0.1284,
       "step": 3750
     },
     {
-      "epoch": 2.915057915057915,
-      "grad_norm": 2.525545358657837,
+      "epoch": 2.3824550331334806,
+      "grad_norm": 2.2240195274353027,
       "learning_rate": 2.7222222222222224e-06,
-      "loss": 0.1534,
+      "loss": 0.1282,
       "step": 3775
     },
     {
-      "epoch": 2.9343629343629343,
-      "grad_norm": 2.6334362030029297,
+      "epoch": 2.398232881035027,
+      "grad_norm": 1.973368525505066,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 0.158,
+      "loss": 0.1261,
       "step": 3800
     },
     {
-      "epoch": 2.9536679536679538,
-      "grad_norm": 2.1491916179656982,
+      "epoch": 2.414010728936573,
+      "grad_norm": 2.7670340538024902,
       "learning_rate": 2.6111111111111113e-06,
-      "loss": 0.15,
+      "loss": 0.1249,
       "step": 3825
     },
     {
-      "epoch": 2.972972972972973,
-      "grad_norm": 2.0909786224365234,
+      "epoch": 2.4297885768381193,
+      "grad_norm": 1.904808759689331,
       "learning_rate": 2.5555555555555557e-06,
-      "loss": 0.147,
+      "loss": 0.1219,
       "step": 3850
     },
     {
-      "epoch": 2.9922779922779923,
-      "grad_norm": 2.8027398586273193,
+      "epoch": 2.4455664247396656,
+      "grad_norm": 2.0226497650146484,
       "learning_rate": 2.5e-06,
-      "loss": 0.1642,
+      "loss": 0.1282,
       "step": 3875
     },
     {
-      "epoch": 3.011583011583012,
-      "grad_norm": 1.8371341228485107,
+      "epoch": 2.461344272641212,
+      "grad_norm": 2.153346300125122,
       "learning_rate": 2.4444444444444447e-06,
-      "loss": 0.1176,
+      "loss": 0.1274,
       "step": 3900
     },
     {
-      "epoch": 3.030888030888031,
-      "grad_norm": 2.1273913383483887,
+      "epoch": 2.477122120542758,
+      "grad_norm": 2.0116922855377197,
       "learning_rate": 2.388888888888889e-06,
-      "loss": 0.0937,
+      "loss": 0.1342,
       "step": 3925
     },
     {
-      "epoch": 3.0501930501930503,
-      "grad_norm": 1.8209782838821411,
+      "epoch": 2.4928999684443043,
+      "grad_norm": 1.9735043048858643,
       "learning_rate": 2.3333333333333336e-06,
-      "loss": 0.089,
+      "loss": 0.1328,
       "step": 3950
     },
     {
-      "epoch": 3.0694980694980694,
-      "grad_norm": 2.118212938308716,
+      "epoch": 2.5086778163458505,
+      "grad_norm": 1.958715796470642,
       "learning_rate": 2.277777777777778e-06,
-      "loss": 0.0888,
+      "loss": 0.1175,
       "step": 3975
     },
     {
-      "epoch": 3.088803088803089,
-      "grad_norm": 1.675663709640503,
+      "epoch": 2.5244556642473968,
+      "grad_norm": 2.146303176879883,
       "learning_rate": 2.222222222222222e-06,
-      "loss": 0.0841,
+      "loss": 0.1308,
       "step": 4000
     },
     {
-      "epoch": 3.088803088803089,
-      "eval_loss": 0.4492855668067932,
-      "eval_runtime": 2954.4178,
-      "eval_samples_per_second": 2.377,
+      "epoch": 2.5244556642473968,
+      "eval_loss": 0.406938761472702,
+      "eval_runtime": 2943.6463,
+      "eval_samples_per_second": 2.385,
       "eval_steps_per_second": 0.149,
-      "eval_wer": 0.2968411212526454,
+      "eval_wer": 0.2964035658770543,
       "step": 4000
     },
     {
-      "epoch": 3.108108108108108,
-      "grad_norm": 2.0060861110687256,
+      "epoch": 2.540233512148943,
+      "grad_norm": 2.378037452697754,
       "learning_rate": 2.166666666666667e-06,
-      "loss": 0.084,
+      "loss": 0.1315,
       "step": 4025
     },
     {
-      "epoch": 3.1274131274131274,
-      "grad_norm": 1.8335351943969727,
+      "epoch": 2.5560113600504892,
+      "grad_norm": 1.4860683679580688,
       "learning_rate": 2.1111111111111114e-06,
-      "loss": 0.0871,
+      "loss": 0.1256,
       "step": 4050
     },
     {
-      "epoch": 3.146718146718147,
-      "grad_norm": 2.165919303894043,
+      "epoch": 2.5717892079520355,
+      "grad_norm": 2.182999849319458,
       "learning_rate": 2.0555555555555555e-06,
-      "loss": 0.0892,
+      "loss": 0.1245,
       "step": 4075
     },
     {
-      "epoch": 3.166023166023166,
-      "grad_norm": 1.5656211376190186,
+      "epoch": 2.5875670558535817,
+      "grad_norm": 2.181810140609741,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0869,
+      "loss": 0.1309,
       "step": 4100
     },
     {
-      "epoch": 3.1853281853281854,
-      "grad_norm": 1.9687210321426392,
+      "epoch": 2.603344903755128,
+      "grad_norm": 2.1204681396484375,
       "learning_rate": 1.944444444444445e-06,
-      "loss": 0.0878,
+      "loss": 0.1308,
       "step": 4125
     },
     {
-      "epoch": 3.2046332046332044,
-      "grad_norm": 2.017395496368408,
+      "epoch": 2.619122751656674,
+      "grad_norm": 2.2567408084869385,
       "learning_rate": 1.888888888888889e-06,
-      "loss": 0.0823,
+      "loss": 0.1292,
       "step": 4150
     },
     {
-      "epoch": 3.223938223938224,
-      "grad_norm": 2.3069615364074707,
-      "learning_rate": 1.8333333333333333e-06,
-      "loss": 0.0894,
+      "epoch": 2.6349005995582204,
+      "grad_norm": 2.3920841217041016,
+      "learning_rate": 1.8355555555555557e-06,
+      "loss": 0.1257,
       "step": 4175
     },
     {
-      "epoch": 3.2432432432432434,
-      "grad_norm": 1.8950186967849731,
-      "learning_rate": 1.777777777777778e-06,
-      "loss": 0.0889,
+      "epoch": 2.650678447459766,
+      "grad_norm": 2.260571241378784,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.1342,
       "step": 4200
     },
     {
-      "epoch": 3.2625482625482625,
-      "grad_norm": 1.9198052883148193,
-      "learning_rate": 1.7222222222222224e-06,
-      "loss": 0.081,
+      "epoch": 2.6664562953613125,
+      "grad_norm": 2.4567618370056152,
+      "learning_rate": 1.7244444444444448e-06,
+      "loss": 0.1279,
       "step": 4225
     },
     {
-      "epoch": 3.281853281853282,
-      "grad_norm": 2.25732421875,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.0872,
+      "epoch": 2.6822341432628587,
+      "grad_norm": 2.1594390869140625,
+      "learning_rate": 1.668888888888889e-06,
+      "loss": 0.1253,
       "step": 4250
     },
     {
-      "epoch": 3.301158301158301,
-      "grad_norm": 1.6695570945739746,
-      "learning_rate": 1.6111111111111113e-06,
-      "loss": 0.0827,
+      "epoch": 2.698011991164405,
+      "grad_norm": 2.029165267944336,
+      "learning_rate": 1.6133333333333335e-06,
+      "loss": 0.128,
       "step": 4275
     },
     {
-      "epoch": 3.3204633204633205,
-      "grad_norm": 2.1860039234161377,
-      "learning_rate": 1.5555555555555558e-06,
-      "loss": 0.0864,
+      "epoch": 2.713789839065951,
+      "grad_norm": 2.011826515197754,
+      "learning_rate": 1.5577777777777777e-06,
+      "loss": 0.1177,
       "step": 4300
     },
     {
-      "epoch": 3.33976833976834,
-      "grad_norm": 2.0618574619293213,
-      "learning_rate": 1.5e-06,
-      "loss": 0.0843,
+      "epoch": 2.7295676869674974,
+      "grad_norm": 2.081116199493408,
+      "learning_rate": 1.5022222222222224e-06,
+      "loss": 0.1322,
       "step": 4325
     },
     {
-      "epoch": 3.359073359073359,
-      "grad_norm": 1.952563762664795,
-      "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.0822,
+      "epoch": 2.7453455348690436,
+      "grad_norm": 1.9539333581924438,
+      "learning_rate": 1.4466666666666669e-06,
+      "loss": 0.1225,
       "step": 4350
     },
     {
-      "epoch": 3.3783783783783785,
-      "grad_norm": 2.2584211826324463,
-      "learning_rate": 1.3888888888888892e-06,
-      "loss": 0.087,
+      "epoch": 2.76112338277059,
+      "grad_norm": 1.83766770362854,
+      "learning_rate": 1.3911111111111111e-06,
+      "loss": 0.1218,
       "step": 4375
     },
     {
-      "epoch": 3.3976833976833976,
-      "grad_norm": 1.7842656373977661,
-      "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.0856,
+      "epoch": 2.776901230672136,
+      "grad_norm": 1.8542882204055786,
+      "learning_rate": 1.3355555555555558e-06,
+      "loss": 0.1215,
       "step": 4400
     },
     {
-      "epoch": 3.416988416988417,
-      "grad_norm": 2.130709409713745,
-      "learning_rate": 1.2777777777777779e-06,
-      "loss": 0.0886,
+      "epoch": 2.7926790785736824,
+      "grad_norm": 1.944549798965454,
+      "learning_rate": 1.28e-06,
+      "loss": 0.1333,
       "step": 4425
     },
     {
-      "epoch": 3.436293436293436,
-      "grad_norm": 1.505071759223938,
-      "learning_rate": 1.2222222222222223e-06,
-      "loss": 0.0795,
+      "epoch": 2.8084569264752286,
+      "grad_norm": 1.9153203964233398,
+      "learning_rate": 1.2244444444444445e-06,
+      "loss": 0.1312,
       "step": 4450
     },
     {
-      "epoch": 3.4555984555984556,
-      "grad_norm": 2.2736964225769043,
-      "learning_rate": 1.1666666666666668e-06,
-      "loss": 0.0802,
+      "epoch": 2.824234774376775,
+      "grad_norm": 1.7963358163833618,
+      "learning_rate": 1.168888888888889e-06,
+      "loss": 0.1202,
       "step": 4475
     },
     {
-      "epoch": 3.474903474903475,
-      "grad_norm": 1.685120940208435,
-      "learning_rate": 1.111111111111111e-06,
-      "loss": 0.085,
+      "epoch": 2.840012622278321,
+      "grad_norm": 2.115541696548462,
+      "learning_rate": 1.1133333333333334e-06,
+      "loss": 0.1234,
       "step": 4500
     },
     {
-      "epoch": 3.494208494208494,
-      "grad_norm": 1.8761180639266968,
-      "learning_rate": 1.0555555555555557e-06,
-      "loss": 0.0732,
+      "epoch": 2.8557904701798673,
+      "grad_norm": 2.2020208835601807,
+      "learning_rate": 1.0577777777777779e-06,
+      "loss": 0.1281,
       "step": 4525
     },
     {
-      "epoch": 3.5135135135135136,
-      "grad_norm": 1.7968302965164185,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0868,
+      "epoch": 2.8715683180814136,
+      "grad_norm": 2.03625226020813,
+      "learning_rate": 1.0022222222222223e-06,
+      "loss": 0.1216,
       "step": 4550
     },
     {
-      "epoch": 3.532818532818533,
-      "grad_norm": 1.6335569620132446,
-      "learning_rate": 9.444444444444445e-07,
-      "loss": 0.0787,
+      "epoch": 2.88734616598296,
+      "grad_norm": 2.0224132537841797,
+      "learning_rate": 9.466666666666667e-07,
+      "loss": 0.1163,
       "step": 4575
     },
     {
-      "epoch": 3.552123552123552,
-      "grad_norm": 1.402917742729187,
-      "learning_rate": 8.88888888888889e-07,
-      "loss": 0.0779,
+      "epoch": 2.903124013884506,
+      "grad_norm": 2.030674695968628,
+      "learning_rate": 8.911111111111112e-07,
+      "loss": 0.1297,
       "step": 4600
     },
     {
-      "epoch": 3.571428571428571,
-      "grad_norm": 1.4878441095352173,
-      "learning_rate": 8.333333333333333e-07,
-      "loss": 0.0795,
+      "epoch": 2.9189018617860523,
+      "grad_norm": 1.9184988737106323,
+      "learning_rate": 8.355555555555556e-07,
+      "loss": 0.1243,
       "step": 4625
     },
     {
-      "epoch": 3.5907335907335907,
-      "grad_norm": 1.846757411956787,
-      "learning_rate": 7.777777777777779e-07,
-      "loss": 0.0868,
+      "epoch": 2.9346797096875985,
+      "grad_norm": 1.922864556312561,
+      "learning_rate": 7.8e-07,
+      "loss": 0.1175,
       "step": 4650
     },
     {
-      "epoch": 3.61003861003861,
-      "grad_norm": 1.496010422706604,
-      "learning_rate": 7.222222222222222e-07,
-      "loss": 0.0836,
+      "epoch": 2.9504575575891447,
+      "grad_norm": 2.3562278747558594,
+      "learning_rate": 7.244444444444446e-07,
+      "loss": 0.1352,
       "step": 4675
     },
     {
-      "epoch": 3.629343629343629,
-      "grad_norm": 2.0897486209869385,
-      "learning_rate": 6.666666666666667e-07,
-      "loss": 0.0839,
+      "epoch": 2.966235405490691,
+      "grad_norm": 1.826926589012146,
+      "learning_rate": 6.68888888888889e-07,
+      "loss": 0.129,
       "step": 4700
     },
     {
-      "epoch": 3.6486486486486487,
-      "grad_norm": 1.6161247491836548,
-      "learning_rate": 6.111111111111112e-07,
-      "loss": 0.0858,
+      "epoch": 2.982013253392237,
+      "grad_norm": 1.7427453994750977,
+      "learning_rate": 6.133333333333333e-07,
+      "loss": 0.1231,
       "step": 4725
     },
     {
-      "epoch": 3.667953667953668,
-      "grad_norm": 1.9887114763259888,
-      "learning_rate": 5.555555555555555e-07,
-      "loss": 0.0864,
+      "epoch": 2.9977911012937835,
+      "grad_norm": 1.7461098432540894,
+      "learning_rate": 5.577777777777779e-07,
+      "loss": 0.1175,
       "step": 4750
     },
     {
-      "epoch": 3.687258687258687,
-      "grad_norm": 2.271867513656616,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0815,
+      "epoch": 3.0135689491953297,
+      "grad_norm": 1.2887001037597656,
+      "learning_rate": 5.022222222222222e-07,
+      "loss": 0.0873,
       "step": 4775
     },
     {
-      "epoch": 3.7065637065637067,
-      "grad_norm": 1.9788146018981934,
-      "learning_rate": 4.444444444444445e-07,
-      "loss": 0.0821,
+      "epoch": 3.029346797096876,
+      "grad_norm": 1.7019718885421753,
+      "learning_rate": 4.466666666666667e-07,
+      "loss": 0.0826,
       "step": 4800
     },
     {
-      "epoch": 3.7258687258687258,
-      "grad_norm": 1.8956130743026733,
-      "learning_rate": 3.8888888888888895e-07,
-      "loss": 0.0789,
+      "epoch": 3.045124644998422,
+      "grad_norm": 1.7305608987808228,
+      "learning_rate": 3.9111111111111115e-07,
+      "loss": 0.0788,
       "step": 4825
     },
     {
-      "epoch": 3.7451737451737452,
-      "grad_norm": 2.4207746982574463,
-      "learning_rate": 3.3333333333333335e-07,
-      "loss": 0.1009,
+      "epoch": 3.0609024928999684,
+      "grad_norm": 1.4667015075683594,
+      "learning_rate": 3.3555555555555556e-07,
+      "loss": 0.0775,
       "step": 4850
     },
     {
-      "epoch": 3.7644787644787643,
-      "grad_norm": 1.7667814493179321,
-      "learning_rate": 2.7777777777777776e-07,
-      "loss": 0.0836,
+      "epoch": 3.0766803408015146,
+      "grad_norm": 2.291264057159424,
+      "learning_rate": 2.8e-07,
+      "loss": 0.0802,
       "step": 4875
     },
     {
-      "epoch": 3.7837837837837838,
-      "grad_norm": 2.118232488632202,
-      "learning_rate": 2.2222222222222224e-07,
-      "loss": 0.0824,
+      "epoch": 3.092458188703061,
+      "grad_norm": 1.8721388578414917,
+      "learning_rate": 2.2444444444444445e-07,
+      "loss": 0.0789,
       "step": 4900
     },
     {
-      "epoch": 3.8030888030888033,
-      "grad_norm": 1.5285857915878296,
-      "learning_rate": 1.6666666666666668e-07,
-      "loss": 0.0836,
+      "epoch": 3.108236036604607,
+      "grad_norm": 1.4734169244766235,
+      "learning_rate": 1.6888888888888888e-07,
+      "loss": 0.0919,
       "step": 4925
     },
     {
-      "epoch": 3.8223938223938223,
-      "grad_norm": 2.041189670562744,
-      "learning_rate": 1.1111111111111112e-07,
-      "loss": 0.0838,
+      "epoch": 3.1240138845061534,
+      "grad_norm": 1.3516019582748413,
+      "learning_rate": 1.1333333333333336e-07,
+      "loss": 0.0775,
       "step": 4950
     },
     {
-      "epoch": 3.841698841698842,
-      "grad_norm": 2.062230110168457,
-      "learning_rate": 5.555555555555556e-08,
-      "loss": 0.0804,
+      "epoch": 3.1397917324076996,
+      "grad_norm": 1.5201116800308228,
+      "learning_rate": 5.777777777777778e-08,
+      "loss": 0.0776,
       "step": 4975
     },
     {
-      "epoch": 3.861003861003861,
-      "grad_norm": 1.9036461114883423,
-      "learning_rate": 0.0,
-      "loss": 0.0845,
+      "epoch": 3.155569580309246,
+      "grad_norm": 1.4460948705673218,
+      "learning_rate": 2.2222222222222225e-09,
+      "loss": 0.0804,
       "step": 5000
     },
     {
-      "epoch": 3.861003861003861,
-      "eval_loss": 0.447899729013443,
-      "eval_runtime": 2978.0577,
-      "eval_samples_per_second": 2.358,
+      "epoch": 3.155569580309246,
+      "eval_loss": 0.4191373586654663,
+      "eval_runtime": 2987.8843,
+      "eval_samples_per_second": 2.35,
       "eval_steps_per_second": 0.147,
-      "eval_wer": 0.29472481531011024,
+      "eval_wer": 0.2910627217048187,
       "step": 5000
     },
     {
-      "epoch": 3.861003861003861,
+      "epoch": 3.155569580309246,
       "step": 5000,
-      "total_flos": 5.435487665750016e+20,
-      "train_loss": 0.2695653033494949,
-      "train_runtime": 64824.1008,
-      "train_samples_per_second": 2.468,
-      "train_steps_per_second": 0.077
+      "total_flos": 5.4352838158516224e+20,
+      "train_loss": 0.2719093270301819,
+      "train_runtime": 63684.9216,
+      "train_samples_per_second": 2.512,
+      "train_steps_per_second": 0.079
     }
   ],
   "logging_steps": 25,
@@ -1480,7 +1480,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.435487665750016e+20,
+  "total_flos": 5.4352838158516224e+20,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null