End of training

README.md

@@ -2,6 +2,8 @@
 license: apache-2.0
 base_model: facebook/wav2vec2-large-xlsr-53
 tags:
+- automatic-speech-recognition
+- DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv
 - generated_from_trainer
 metrics:
 - wer
@@ -15,7 +17,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # wav2vec2-xlsr-53-ft-btb-ccv-cy
 
-This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on an unknown dataset.
+This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the DEWIBRYNJONES/BANC-TRAWSGRIFIADAU-BANGOR-CLEAN-WITH-CCV - DEFAULT dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.7622
 - Wer: 0.5187

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 1.0080645161290323,
-    "eval_loss": 0.5417820811271667,
-    "eval_runtime": 188.0972,
+    "epoch": 1.6803898504453034,
+    "eval_loss": 0.7622119784355164,
+    "eval_runtime": 181.0985,
     "eval_samples": 7022,
-    "eval_samples_per_second": 37.332,
-    "eval_steps_per_second": 2.334,
-    "eval_wer": 0.4178370693206128,
-    "total_flos": 6.709869296482936e+18,
-    "train_loss": 1.3100875549316406,
-    "train_runtime": 4950.9591,
+    "eval_samples_per_second": 38.774,
+    "eval_steps_per_second": 2.424,
+    "eval_wer": 0.518741944485045,
+    "total_flos": 1.1109120866520904e+19,
+    "train_loss": 0.9622068939208984,
+    "train_runtime": 7532.6917,
     "train_samples": 47607,
-    "train_samples_per_second": 9.695,
-    "train_steps_per_second": 0.606
+    "train_samples_per_second": 10.62,
+    "train_steps_per_second": 1.328
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 1.0080645161290323,
-    "eval_loss": 0.5417820811271667,
-    "eval_runtime": 188.0972,
+    "epoch": 1.6803898504453034,
+    "eval_loss": 0.7622119784355164,
+    "eval_runtime": 181.0985,
     "eval_samples": 7022,
-    "eval_samples_per_second": 37.332,
-    "eval_steps_per_second": 2.334,
-    "eval_wer": 0.4178370693206128
+    "eval_samples_per_second": 38.774,
+    "eval_steps_per_second": 2.424,
+    "eval_wer": 0.518741944485045
 }

runs/Aug29_22-44-30_6d77da852b30/events.out.tfevents.1724976520.6d77da852b30.1954.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19f8712d7f4a903ff1562719ce4868b4e22853555b545994fdd7895e59221dce
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 1.0080645161290323,
-    "total_flos": 6.709869296482936e+18,
-    "train_loss": 1.3100875549316406,
-    "train_runtime": 4950.9591,
+    "epoch": 1.6803898504453034,
+    "total_flos": 1.1109120866520904e+19,
+    "train_loss": 0.9622068939208984,
+    "train_runtime": 7532.6917,
     "train_samples": 47607,
-    "train_samples_per_second": 9.695,
-    "train_steps_per_second": 0.606
+    "train_samples_per_second": 10.62,
+    "train_steps_per_second": 1.328
 }

trainer_state.json

@@ -1,202 +1,345 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0080645161290323,
-  "eval_steps": 200,
-  "global_step": 3000,
+  "epoch": 1.6803898504453034,
+  "eval_steps": 500,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.06720430107526881,
-      "eval_loss": 3.1445324420928955,
-      "eval_runtime": 187.2734,
-      "eval_samples_per_second": 37.496,
-      "eval_steps_per_second": 2.344,
-      "eval_wer": 1.0,
-      "step": 200
+      "epoch": 0.08401949252226516,
+      "grad_norm": 4.783174514770508,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 4.5088,
+      "step": 500
     },
     {
-      "epoch": 0.13440860215053763,
-      "eval_loss": 2.7407212257385254,
-      "eval_runtime": 185.5858,
-      "eval_samples_per_second": 37.837,
-      "eval_steps_per_second": 2.365,
-      "eval_wer": 0.9999867125526515,
-      "step": 400
+      "epoch": 0.08401949252226516,
+      "eval_loss": 2.0263099670410156,
+      "eval_runtime": 176.3381,
+      "eval_samples_per_second": 39.821,
+      "eval_steps_per_second": 2.49,
+      "eval_wer": 1.0069493349632601,
+      "step": 500
     },
     {
-      "epoch": 0.16801075268817203,
-      "grad_norm": 8.384466171264648,
-      "learning_rate": 0.00027866666666666665,
-      "loss": 4.0188,
-      "step": 500
+      "epoch": 0.16803898504453033,
+      "grad_norm": 5.4134521484375,
+      "learning_rate": 0.0002844631578947368,
+      "loss": 1.2959,
+      "step": 1000
     },
     {
-      "epoch": 0.20161290322580644,
-      "eval_loss": 1.2700377702713013,
-      "eval_runtime": 187.6926,
-      "eval_samples_per_second": 37.412,
-      "eval_steps_per_second": 2.339,
-      "eval_wer": 0.8484433755431244,
-      "step": 600
+      "epoch": 0.16803898504453033,
+      "eval_loss": 1.0947818756103516,
+      "eval_runtime": 175.5242,
+      "eval_samples_per_second": 40.006,
+      "eval_steps_per_second": 2.501,
+      "eval_wer": 0.7472196016423285,
+      "step": 1000
     },
     {
-      "epoch": 0.26881720430107525,
-      "eval_loss": 0.9953192472457886,
-      "eval_runtime": 187.642,
-      "eval_samples_per_second": 37.422,
-      "eval_steps_per_second": 2.34,
-      "eval_wer": 0.7435389787267968,
-      "step": 800
+      "epoch": 0.2520584775667955,
+      "grad_norm": 3.8002684116363525,
+      "learning_rate": 0.00026867368421052627,
+      "loss": 0.9954,
+      "step": 1500
     },
     {
-      "epoch": 0.33602150537634407,
-      "grad_norm": 5.077725410461426,
-      "learning_rate": 0.0002232222222222222,
-      "loss": 1.0707,
-      "step": 1000
+      "epoch": 0.2520584775667955,
+      "eval_loss": 0.9248552918434143,
+      "eval_runtime": 177.606,
+      "eval_samples_per_second": 39.537,
+      "eval_steps_per_second": 2.472,
+      "eval_wer": 0.6713615647297998,
+      "step": 1500
     },
     {
-      "epoch": 0.33602150537634407,
-      "eval_loss": 0.8646696209907532,
-      "eval_runtime": 187.682,
-      "eval_samples_per_second": 37.414,
-      "eval_steps_per_second": 2.339,
-      "eval_wer": 0.6541277455188084,
-      "step": 1000
+      "epoch": 0.33607797008906065,
+      "grad_norm": 6.10047721862793,
+      "learning_rate": 0.0002529157894736842,
+      "loss": 0.8907,
+      "step": 2000
     },
     {
-      "epoch": 0.4032258064516129,
-      "eval_loss": 0.7888585925102234,
-      "eval_runtime": 187.1558,
-      "eval_samples_per_second": 37.52,
-      "eval_steps_per_second": 2.346,
-      "eval_wer": 0.5784025830797646,
-      "step": 1200
+      "epoch": 0.33607797008906065,
+      "eval_loss": 0.8197005987167358,
+      "eval_runtime": 176.9088,
+      "eval_samples_per_second": 39.693,
+      "eval_steps_per_second": 2.482,
+      "eval_wer": 0.6131093955540201,
+      "step": 2000
     },
     {
-      "epoch": 0.47043010752688175,
-      "eval_loss": 0.7465152740478516,
-      "eval_runtime": 185.9206,
-      "eval_samples_per_second": 37.769,
-      "eval_steps_per_second": 2.361,
-      "eval_wer": 0.5440412442365697,
-      "step": 1400
+      "epoch": 0.4200974626113258,
+      "grad_norm": 7.82942533493042,
+      "learning_rate": 0.00023715789473684206,
+      "loss": 0.8197,
+      "step": 2500
     },
     {
-      "epoch": 0.5040322580645161,
-      "grad_norm": 7.621553897857666,
-      "learning_rate": 0.00016777777777777776,
-      "loss": 0.8175,
-      "step": 1500
+      "epoch": 0.4200974626113258,
+      "eval_loss": 0.7610413432121277,
+      "eval_runtime": 176.4046,
+      "eval_samples_per_second": 39.806,
+      "eval_steps_per_second": 2.489,
+      "eval_wer": 0.550126895122178,
+      "step": 2500
+    },
+    {
+      "epoch": 0.504116955133591,
+      "grad_norm": 7.377211093902588,
+      "learning_rate": 0.00022136842105263156,
+      "loss": 0.7873,
+      "step": 3000
     },
     {
-      "epoch": 0.5376344086021505,
-      "eval_loss": 0.68277907371521,
-      "eval_runtime": 187.6331,
-      "eval_samples_per_second": 37.424,
-      "eval_steps_per_second": 2.34,
-      "eval_wer": 0.5042453394278426,
-      "step": 1600
+      "epoch": 0.504116955133591,
+      "eval_loss": 0.7495535612106323,
+      "eval_runtime": 176.1708,
+      "eval_samples_per_second": 39.859,
+      "eval_steps_per_second": 2.492,
+      "eval_wer": 0.5436824831581605,
+      "step": 3000
     },
     {
-      "epoch": 0.6048387096774194,
-      "eval_loss": 0.6549283862113953,
-      "eval_runtime": 186.6705,
-      "eval_samples_per_second": 37.617,
-      "eval_steps_per_second": 2.352,
-      "eval_wer": 0.4952098752308694,
-      "step": 1800
+      "epoch": 0.5881364476558562,
+      "grad_norm": 7.327758312225342,
+      "learning_rate": 0.00020557894736842103,
+      "loss": 0.748,
+      "step": 3500
     },
     {
-      "epoch": 0.6720430107526881,
-      "grad_norm": 6.983826637268066,
-      "learning_rate": 0.00011233333333333333,
-      "loss": 0.7148,
-      "step": 2000
+      "epoch": 0.5881364476558562,
+      "eval_loss": 0.6833189725875854,
+      "eval_runtime": 175.5214,
+      "eval_samples_per_second": 40.007,
+      "eval_steps_per_second": 2.501,
+      "eval_wer": 0.52397719874035,
+      "step": 3500
     },
     {
-      "epoch": 0.6720430107526881,
-      "eval_loss": 0.6289859414100647,
-      "eval_runtime": 188.3084,
-      "eval_samples_per_second": 37.29,
-      "eval_steps_per_second": 2.331,
-      "eval_wer": 0.4905858435535949,
-      "step": 2000
+      "epoch": 0.6721559401781213,
+      "grad_norm": 6.962770938873291,
+      "learning_rate": 0.00018982105263157893,
+      "loss": 0.7133,
+      "step": 4000
     },
     {
-      "epoch": 0.739247311827957,
-      "eval_loss": 0.6112708449363708,
-      "eval_runtime": 188.0636,
-      "eval_samples_per_second": 37.338,
-      "eval_steps_per_second": 2.334,
-      "eval_wer": 0.45763297412934,
-      "step": 2200
+      "epoch": 0.6721559401781213,
+      "eval_loss": 0.6442425847053528,
+      "eval_runtime": 176.4204,
+      "eval_samples_per_second": 39.803,
+      "eval_steps_per_second": 2.488,
+      "eval_wer": 0.4789859020183633,
+      "step": 4000
     },
     {
-      "epoch": 0.8064516129032258,
-      "eval_loss": 0.5718730688095093,
-      "eval_runtime": 187.8424,
-      "eval_samples_per_second": 37.382,
-      "eval_steps_per_second": 2.337,
-      "eval_wer": 0.4404788796024396,
-      "step": 2400
+      "epoch": 0.7561754327003865,
+      "grad_norm": 27.43330955505371,
+      "learning_rate": 0.00017406315789473683,
+      "loss": 0.6809,
+      "step": 4500
     },
     {
-      "epoch": 0.8400537634408602,
-      "grad_norm": 6.760587215423584,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 0.6374,
-      "step": 2500
+      "epoch": 0.7561754327003865,
+      "eval_loss": 0.6224470734596252,
+      "eval_runtime": 176.8219,
+      "eval_samples_per_second": 39.712,
+      "eval_steps_per_second": 2.483,
+      "eval_wer": 0.47034906124184483,
+      "step": 4500
     },
     {
-      "epoch": 0.8736559139784946,
-      "eval_loss": 0.5643858313560486,
-      "eval_runtime": 188.0058,
-      "eval_samples_per_second": 37.35,
-      "eval_steps_per_second": 2.335,
-      "eval_wer": 0.431376978168724,
-      "step": 2600
+      "epoch": 0.8401949252226516,
+      "grad_norm": 6.05275297164917,
+      "learning_rate": 0.00015830526315789472,
+      "loss": 0.6517,
+      "step": 5000
     },
     {
-      "epoch": 0.9408602150537635,
-      "eval_loss": 0.5482733249664307,
-      "eval_runtime": 188.1104,
-      "eval_samples_per_second": 37.329,
-      "eval_steps_per_second": 2.334,
-      "eval_wer": 0.41896650234523447,
-      "step": 2800
+      "epoch": 0.8401949252226516,
+      "eval_loss": 0.599228024482727,
+      "eval_runtime": 177.562,
+      "eval_samples_per_second": 39.547,
+      "eval_steps_per_second": 2.472,
+      "eval_wer": 0.45710147623540043,
+      "step": 5000
     },
     {
-      "epoch": 1.0080645161290323,
-      "grad_norm": 1.7233390808105469,
-      "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.6013,
-      "step": 3000
+      "epoch": 0.9242144177449169,
+      "grad_norm": 6.2337541580200195,
+      "learning_rate": 0.0001425157894736842,
+      "loss": 0.6363,
+      "step": 5500
     },
     {
-      "epoch": 1.0080645161290323,
-      "eval_loss": 0.5417820811271667,
-      "eval_runtime": 189.023,
-      "eval_samples_per_second": 37.149,
-      "eval_steps_per_second": 2.322,
-      "eval_wer": 0.4178370693206128,
-      "step": 3000
+      "epoch": 0.9242144177449169,
+      "eval_loss": 0.5727323293685913,
+      "eval_runtime": 176.4451,
+      "eval_samples_per_second": 39.797,
+      "eval_steps_per_second": 2.488,
+      "eval_wer": 0.4373297545808475,
+      "step": 5500
+    },
+    {
+      "epoch": 1.008233910267182,
+      "grad_norm": 9.131902694702148,
+      "learning_rate": 0.0001267263157894737,
+      "loss": 0.6111,
+      "step": 6000
+    },
+    {
+      "epoch": 1.008233910267182,
+      "eval_loss": 0.5573195219039917,
+      "eval_runtime": 177.5647,
+      "eval_samples_per_second": 39.546,
+      "eval_steps_per_second": 2.472,
+      "eval_wer": 0.42004278558046215,
+      "step": 6000
+    },
+    {
+      "epoch": 1.092253402789447,
+      "grad_norm": 7.859574794769287,
+      "learning_rate": 0.00011096842105263158,
+      "loss": 0.5352,
+      "step": 6500
+    },
+    {
+      "epoch": 1.092253402789447,
+      "eval_loss": 0.5566655993461609,
+      "eval_runtime": 176.9291,
+      "eval_samples_per_second": 39.688,
+      "eval_steps_per_second": 2.481,
+      "eval_wer": 0.4258892624137977,
+      "step": 6500
+    },
+    {
+      "epoch": 1.1762728953117123,
+      "grad_norm": 10.820630073547363,
+      "learning_rate": 9.521052631578946e-05,
+      "loss": 0.6231,
+      "step": 7000
+    },
+    {
+      "epoch": 1.1762728953117123,
+      "eval_loss": 0.825137734413147,
+      "eval_runtime": 177.7828,
+      "eval_samples_per_second": 39.498,
+      "eval_steps_per_second": 2.469,
+      "eval_wer": 0.6299844536866023,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2602923878339776,
+      "grad_norm": 11.600907325744629,
+      "learning_rate": 7.942105263157894e-05,
+      "loss": 0.7995,
+      "step": 7500
+    },
+    {
+      "epoch": 1.2602923878339776,
+      "eval_loss": 0.660980761051178,
+      "eval_runtime": 178.3219,
+      "eval_samples_per_second": 39.378,
+      "eval_steps_per_second": 2.462,
+      "eval_wer": 0.4686482679812381,
+      "step": 7500
+    },
+    {
+      "epoch": 1.3443118803562426,
+      "grad_norm": 35.45184326171875,
+      "learning_rate": 6.363157894736841e-05,
+      "loss": 0.7696,
+      "step": 8000
+    },
+    {
+      "epoch": 1.3443118803562426,
+      "eval_loss": 0.733095109462738,
+      "eval_runtime": 178.388,
+      "eval_samples_per_second": 39.364,
+      "eval_steps_per_second": 2.461,
+      "eval_wer": 0.4870380951115481,
+      "step": 8000
+    },
+    {
+      "epoch": 1.4283313728785079,
+      "grad_norm": 3.933194875717163,
+      "learning_rate": 4.784210526315789e-05,
+      "loss": 0.7978,
+      "step": 8500
+    },
+    {
+      "epoch": 1.4283313728785079,
+      "eval_loss": 0.7790142297744751,
+      "eval_runtime": 178.093,
+      "eval_samples_per_second": 39.429,
+      "eval_steps_per_second": 2.465,
+      "eval_wer": 0.5432041350536149,
+      "step": 8500
+    },
+    {
+      "epoch": 1.512350865400773,
+      "grad_norm": 16.54388427734375,
+      "learning_rate": 3.208421052631579e-05,
+      "loss": 0.7844,
+      "step": 9000
+    },
+    {
+      "epoch": 1.512350865400773,
+      "eval_loss": 0.8260045647621155,
+      "eval_runtime": 177.7102,
+      "eval_samples_per_second": 39.514,
+      "eval_steps_per_second": 2.47,
+      "eval_wer": 0.5481603529146015,
+      "step": 9000
+    },
+    {
+      "epoch": 1.5963703579230382,
+      "grad_norm": 179.12344360351562,
+      "learning_rate": 1.632631578947368e-05,
+      "loss": 0.8211,
+      "step": 9500
+    },
+    {
+      "epoch": 1.5963703579230382,
+      "eval_loss": 0.8065445423126221,
+      "eval_runtime": 177.2427,
+      "eval_samples_per_second": 39.618,
+      "eval_steps_per_second": 2.477,
+      "eval_wer": 0.5491303365710414,
+      "step": 9500
+    },
+    {
+      "epoch": 1.6803898504453034,
+      "grad_norm": 22.516687393188477,
+      "learning_rate": 5.684210526315788e-07,
+      "loss": 0.7744,
+      "step": 10000
+    },
+    {
+      "epoch": 1.6803898504453034,
+      "eval_loss": 0.7622119784355164,
+      "eval_runtime": 177.7095,
+      "eval_samples_per_second": 39.514,
+      "eval_steps_per_second": 2.47,
+      "eval_wer": 0.518741944485045,
+      "step": 10000
     },
     {
-      "epoch": 1.0080645161290323,
-      "step": 3000,
-      "total_flos": 6.709869296482936e+18,
-      "train_loss": 1.3100875549316406,
-      "train_runtime": 4950.9591,
-      "train_samples_per_second": 9.695,
-      "train_steps_per_second": 0.606
+      "epoch": 1.6803898504453034,
+      "step": 10000,
+      "total_flos": 1.1109120866520904e+19,
+      "train_loss": 0.9622068939208984,
+      "train_runtime": 7532.6917,
+      "train_samples_per_second": 10.62,
+      "train_steps_per_second": 1.328
     }
   ],
   "logging_steps": 500,
-  "max_steps": 3000,
+  "max_steps": 10000,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
   "save_steps": 500,
@@ -212,8 +355,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 6.709869296482936e+18,
-  "train_batch_size": 16,
+  "total_flos": 1.1109120866520904e+19,
+  "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null
 }