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: 0.4942
 - Wer: 0.3917

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 2.0161290322580645,
-    "eval_loss": 1.8412970304489136,
-    "eval_runtime": 190.5883,
+    "epoch": 1.0080645161290323,
+    "eval_loss": 0.4942198395729065,
+    "eval_runtime": 177.2206,
     "eval_samples": 7022,
-    "eval_samples_per_second": 36.844,
-    "eval_steps_per_second": 0.577,
-    "eval_wer": 0.9956417172696953,
-    "total_flos": 1.3379687773166643e+19,
-    "train_loss": 1.3496905517578126,
-    "train_runtime": 9965.7858,
+    "eval_samples_per_second": 39.623,
+    "eval_steps_per_second": 2.477,
+    "eval_wer": 0.3917272352808302,
+    "total_flos": 6.709869296482936e+18,
+    "train_loss": 1.1787635701497396,
+    "train_runtime": 4443.9077,
     "train_samples": 47607,
-    "train_samples_per_second": 9.633,
-    "train_steps_per_second": 0.602
+    "train_samples_per_second": 10.801,
+    "train_steps_per_second": 0.675
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.0161290322580645,
-    "eval_loss": 1.8412970304489136,
-    "eval_runtime": 190.5883,
+    "epoch": 1.0080645161290323,
+    "eval_loss": 0.4942198395729065,
+    "eval_runtime": 177.2206,
     "eval_samples": 7022,
-    "eval_samples_per_second": 36.844,
-    "eval_steps_per_second": 0.577,
-    "eval_wer": 0.9956417172696953
+    "eval_samples_per_second": 39.623,
+    "eval_steps_per_second": 2.477,
+    "eval_wer": 0.3917272352808302
 }

runs/Aug29_18-19-07_6d77da852b30/events.out.tfevents.1724957478.6d77da852b30.1440.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bcfb656ce16f70f93fd347aed2079d1fe51fb54b44d46b0ebc76ad7d88eba1e
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.0161290322580645,
-    "total_flos": 1.3379687773166643e+19,
-    "train_loss": 1.3496905517578126,
-    "train_runtime": 9965.7858,
+    "epoch": 1.0080645161290323,
+    "total_flos": 6.709869296482936e+18,
+    "train_loss": 1.1787635701497396,
+    "train_runtime": 4443.9077,
     "train_samples": 47607,
-    "train_samples_per_second": 9.633,
-    "train_steps_per_second": 0.602
+    "train_samples_per_second": 10.801,
+    "train_steps_per_second": 0.675
 }

trainer_state.json

@@ -1,381 +1,204 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.0161290322580645,
+  "epoch": 1.0080645161290323,
   "eval_steps": 200,
-  "global_step": 6000,
+  "global_step": 3000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.06720430107526881,
-      "eval_loss": 3.080493211746216,
-      "eval_runtime": 198.5178,
-      "eval_samples_per_second": 35.372,
-      "eval_steps_per_second": 0.554,
+      "eval_loss": 2.9829530715942383,
+      "eval_runtime": 176.5397,
+      "eval_samples_per_second": 39.776,
+      "eval_steps_per_second": 2.487,
       "eval_wer": 1.0,
       "step": 200
     },
     {
       "epoch": 0.13440860215053763,
-      "eval_loss": 2.92264461517334,
-      "eval_runtime": 201.3326,
-      "eval_samples_per_second": 34.878,
-      "eval_steps_per_second": 0.546,
-      "eval_wer": 1.0,
+      "eval_loss": 1.5613375902175903,
+      "eval_runtime": 174.5758,
+      "eval_samples_per_second": 40.223,
+      "eval_steps_per_second": 2.515,
+      "eval_wer": 0.9657582481829415,
       "step": 400
     },
     {
       "epoch": 0.16801075268817203,
-      "grad_norm": 5.131893634796143,
-      "learning_rate": 0.00024599999999999996,
-      "loss": 4.4646,
+      "grad_norm": 2.138428211212158,
+      "learning_rate": 0.00027833333333333334,
+      "loss": 3.6118,
       "step": 500
     },
     {
       "epoch": 0.20161290322580644,
-      "eval_loss": 1.3392444849014282,
-      "eval_runtime": 198.1589,
-      "eval_samples_per_second": 35.436,
-      "eval_steps_per_second": 0.555,
-      "eval_wer": 0.8646141989662366,
+      "eval_loss": 1.0701438188552856,
+      "eval_runtime": 174.6767,
+      "eval_samples_per_second": 40.2,
+      "eval_steps_per_second": 2.513,
+      "eval_wer": 0.7648786191684716,
       "step": 600
     },
     {
       "epoch": 0.26881720430107525,
-      "eval_loss": 1.0115001201629639,
-      "eval_runtime": 200.6717,
-      "eval_samples_per_second": 34.992,
-      "eval_steps_per_second": 0.548,
-      "eval_wer": 0.7168179221089837,
+      "eval_loss": 0.8867517113685608,
+      "eval_runtime": 174.8944,
+      "eval_samples_per_second": 40.15,
+      "eval_steps_per_second": 2.51,
+      "eval_wer": 0.6947474720631419,
       "step": 800
     },
     {
       "epoch": 0.33602150537634407,
-      "grad_norm": 6.313941478729248,
-      "learning_rate": 0.0002782777777777778,
-      "loss": 1.1042,
+      "grad_norm": 2.852606773376465,
+      "learning_rate": 0.00022288888888888887,
+      "loss": 0.9333,
       "step": 1000
     },
     {
       "epoch": 0.33602150537634407,
-      "eval_loss": 0.9008844494819641,
-      "eval_runtime": 199.6418,
-      "eval_samples_per_second": 35.173,
-      "eval_steps_per_second": 0.551,
-      "eval_wer": 0.6766765436691957,
+      "eval_loss": 0.7679557204246521,
+      "eval_runtime": 175.5538,
+      "eval_samples_per_second": 39.999,
+      "eval_steps_per_second": 2.501,
+      "eval_wer": 0.6070503195631087,
       "step": 1000
     },
     {
       "epoch": 0.4032258064516129,
-      "eval_loss": 0.8225907683372498,
-      "eval_runtime": 199.7483,
-      "eval_samples_per_second": 35.154,
-      "eval_steps_per_second": 0.551,
-      "eval_wer": 0.5947727182131041,
+      "eval_loss": 0.7223904132843018,
+      "eval_runtime": 175.6961,
+      "eval_samples_per_second": 39.967,
+      "eval_steps_per_second": 2.499,
+      "eval_wer": 0.5453965638661157,
       "step": 1200
     },
     {
       "epoch": 0.47043010752688175,
-      "eval_loss": 0.8068214058876038,
-      "eval_runtime": 197.9234,
-      "eval_samples_per_second": 35.478,
-      "eval_steps_per_second": 0.556,
-      "eval_wer": 0.5977358189718174,
+      "eval_loss": 0.6732765436172485,
+      "eval_runtime": 175.6064,
+      "eval_samples_per_second": 39.987,
+      "eval_steps_per_second": 2.5,
+      "eval_wer": 0.5121646580475425,
       "step": 1400
     },
     {
       "epoch": 0.5040322580645161,
-      "grad_norm": 8.026273727416992,
-      "learning_rate": 0.0002505555555555555,
-      "loss": 0.8421,
+      "grad_norm": 4.027652740478516,
+      "learning_rate": 0.00016744444444444443,
+      "loss": 0.7446,
       "step": 1500
     },
     {
       "epoch": 0.5376344086021505,
-      "eval_loss": 0.7186344861984253,
-      "eval_runtime": 197.8905,
-      "eval_samples_per_second": 35.484,
-      "eval_steps_per_second": 0.556,
-      "eval_wer": 0.5375171075884612,
+      "eval_loss": 0.6437448859214783,
+      "eval_runtime": 175.5304,
+      "eval_samples_per_second": 40.004,
+      "eval_steps_per_second": 2.501,
+      "eval_wer": 0.4966449195445063,
       "step": 1600
     },
     {
       "epoch": 0.6048387096774194,
-      "eval_loss": 0.7200700640678406,
-      "eval_runtime": 197.0513,
-      "eval_samples_per_second": 35.635,
-      "eval_steps_per_second": 0.558,
-      "eval_wer": 0.5560796715343016,
+      "eval_loss": 0.6063565015792847,
+      "eval_runtime": 175.877,
+      "eval_samples_per_second": 39.926,
+      "eval_steps_per_second": 2.496,
+      "eval_wer": 0.4774312706785899,
       "step": 1800
     },
     {
       "epoch": 0.6720430107526881,
-      "grad_norm": 4.496196746826172,
-      "learning_rate": 0.00022283333333333332,
-      "loss": 0.7637,
+      "grad_norm": 3.676745653152466,
+      "learning_rate": 0.000112,
+      "loss": 0.6579,
       "step": 2000
     },
     {
       "epoch": 0.6720430107526881,
-      "eval_loss": 0.6747044920921326,
-      "eval_runtime": 198.0259,
-      "eval_samples_per_second": 35.46,
-      "eval_steps_per_second": 0.555,
-      "eval_wer": 0.5086966342895867,
+      "eval_loss": 0.5673760771751404,
+      "eval_runtime": 176.4608,
+      "eval_samples_per_second": 39.794,
+      "eval_steps_per_second": 2.488,
+      "eval_wer": 0.44613933217289625,
       "step": 2000
     },
     {
       "epoch": 0.739247311827957,
-      "eval_loss": 0.6709386706352234,
-      "eval_runtime": 199.0402,
-      "eval_samples_per_second": 35.279,
-      "eval_steps_per_second": 0.553,
-      "eval_wer": 0.5050558737161004,
+      "eval_loss": 0.5556111931800842,
+      "eval_runtime": 175.9454,
+      "eval_samples_per_second": 39.91,
+      "eval_steps_per_second": 2.495,
+      "eval_wer": 0.4325196986406941,
       "step": 2200
     },
     {
       "epoch": 0.8064516129032258,
-      "eval_loss": 0.6484177708625793,
-      "eval_runtime": 193.7002,
-      "eval_samples_per_second": 36.252,
-      "eval_steps_per_second": 0.568,
-      "eval_wer": 0.48504497800927465,
+      "eval_loss": 0.5264282822608948,
+      "eval_runtime": 176.4778,
+      "eval_samples_per_second": 39.79,
+      "eval_steps_per_second": 2.488,
+      "eval_wer": 0.4180363810308402,
       "step": 2400
     },
     {
       "epoch": 0.8400537634408602,
-      "grad_norm": 4.925416469573975,
-      "learning_rate": 0.00019516666666666665,
-      "loss": 0.7134,
+      "grad_norm": 2.7548441886901855,
+      "learning_rate": 5.666666666666666e-05,
+      "loss": 0.5823,
       "step": 2500
     },
     {
       "epoch": 0.8736559139784946,
-      "eval_loss": 0.6387702226638794,
-      "eval_runtime": 191.5475,
-      "eval_samples_per_second": 36.659,
-      "eval_steps_per_second": 0.574,
-      "eval_wer": 0.4915159648679892,
+      "eval_loss": 0.5129852890968323,
+      "eval_runtime": 176.2054,
+      "eval_samples_per_second": 39.851,
+      "eval_steps_per_second": 2.491,
+      "eval_wer": 0.4022110312387887,
       "step": 2600
     },
     {
       "epoch": 0.9408602150537635,
-      "eval_loss": 0.627938985824585,
-      "eval_runtime": 192.2841,
-      "eval_samples_per_second": 36.519,
-      "eval_steps_per_second": 0.572,
-      "eval_wer": 0.4705217980573752,
+      "eval_loss": 0.49821802973747253,
+      "eval_runtime": 176.2887,
+      "eval_samples_per_second": 39.832,
+      "eval_steps_per_second": 2.49,
+      "eval_wer": 0.39360076535696725,
       "step": 2800
     },
     {
       "epoch": 1.0080645161290323,
-      "grad_norm": 2.991081714630127,
-      "learning_rate": 0.00016738888888888888,
-      "loss": 0.6968,
+      "grad_norm": 0.7031016945838928,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.5426,
       "step": 3000
     },
     {
       "epoch": 1.0080645161290323,
-      "eval_loss": 0.6451985836029053,
-      "eval_runtime": 192.2109,
-      "eval_samples_per_second": 36.533,
-      "eval_steps_per_second": 0.572,
-      "eval_wer": 0.4833441847486679,
+      "eval_loss": 0.4942198395729065,
+      "eval_runtime": 177.2018,
+      "eval_samples_per_second": 39.627,
+      "eval_steps_per_second": 2.477,
+      "eval_wer": 0.3917272352808302,
       "step": 3000
     },
     {
-      "epoch": 1.075268817204301,
-      "eval_loss": 0.6439228057861328,
-      "eval_runtime": 191.7283,
-      "eval_samples_per_second": 36.625,
-      "eval_steps_per_second": 0.574,
-      "eval_wer": 0.4896955845812461,
-      "step": 3200
-    },
-    {
-      "epoch": 1.14247311827957,
-      "eval_loss": 0.6749800443649292,
-      "eval_runtime": 191.0444,
-      "eval_samples_per_second": 36.756,
-      "eval_steps_per_second": 0.576,
-      "eval_wer": 0.5010164897221595,
-      "step": 3400
-    },
-    {
-      "epoch": 1.1760752688172043,
-      "grad_norm": 17.283771514892578,
-      "learning_rate": 0.00013977777777777776,
-      "loss": 0.667,
-      "step": 3500
-    },
-    {
-      "epoch": 1.2096774193548387,
-      "eval_loss": 0.7904173135757446,
-      "eval_runtime": 191.7132,
-      "eval_samples_per_second": 36.628,
-      "eval_steps_per_second": 0.574,
-      "eval_wer": 0.5591756467664997,
-      "step": 3600
-    },
-    {
-      "epoch": 1.2768817204301075,
-      "eval_loss": 0.9479840993881226,
-      "eval_runtime": 192.8834,
-      "eval_samples_per_second": 36.405,
-      "eval_steps_per_second": 0.57,
-      "eval_wer": 0.5800502265509773,
-      "step": 3800
-    },
-    {
-      "epoch": 1.3440860215053765,
-      "grad_norm": 26.34053611755371,
-      "learning_rate": 0.0001121111111111111,
-      "loss": 0.9326,
-      "step": 4000
-    },
-    {
-      "epoch": 1.3440860215053765,
-      "eval_loss": 1.1457170248031616,
-      "eval_runtime": 189.9612,
-      "eval_samples_per_second": 36.965,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.6581804169600978,
-      "step": 4000
-    },
-    {
-      "epoch": 1.4112903225806452,
-      "eval_loss": 1.118348479270935,
-      "eval_runtime": 190.0092,
-      "eval_samples_per_second": 36.956,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.6742183659097251,
-      "step": 4200
-    },
-    {
-      "epoch": 1.478494623655914,
-      "eval_loss": 1.3061094284057617,
-      "eval_runtime": 190.4579,
-      "eval_samples_per_second": 36.869,
-      "eval_steps_per_second": 0.578,
-      "eval_wer": 0.78073054385522,
-      "step": 4400
-    },
-    {
-      "epoch": 1.5120967741935485,
-      "grad_norm": 15.129014015197754,
-      "learning_rate": 8.433333333333331e-05,
-      "loss": 1.2308,
-      "step": 4500
-    },
-    {
-      "epoch": 1.5456989247311828,
-      "eval_loss": 1.2298626899719238,
-      "eval_runtime": 190.6991,
-      "eval_samples_per_second": 36.822,
-      "eval_steps_per_second": 0.577,
-      "eval_wer": 0.7672304973491543,
-      "step": 4600
-    },
-    {
-      "epoch": 1.6129032258064515,
-      "eval_loss": 1.178423523902893,
-      "eval_runtime": 190.1804,
-      "eval_samples_per_second": 36.923,
-      "eval_steps_per_second": 0.578,
-      "eval_wer": 0.7516974713987696,
-      "step": 4800
-    },
-    {
-      "epoch": 1.6801075268817205,
-      "grad_norm": 3.6261146068573,
-      "learning_rate": 5.6555555555555555e-05,
-      "loss": 1.2835,
-      "step": 5000
-    },
-    {
-      "epoch": 1.6801075268817205,
-      "eval_loss": 1.339849829673767,
-      "eval_runtime": 189.6315,
-      "eval_samples_per_second": 37.03,
-      "eval_steps_per_second": 0.58,
-      "eval_wer": 0.7625267409877888,
-      "step": 5000
-    },
-    {
-      "epoch": 1.7473118279569892,
-      "eval_loss": 1.5754368305206299,
-      "eval_runtime": 190.0111,
-      "eval_samples_per_second": 36.956,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.895573951288218,
-      "step": 5200
-    },
-    {
-      "epoch": 1.814516129032258,
-      "eval_loss": 1.728023886680603,
-      "eval_runtime": 190.3066,
-      "eval_samples_per_second": 36.898,
-      "eval_steps_per_second": 0.578,
-      "eval_wer": 0.9705550166757464,
-      "step": 5400
-    },
-    {
-      "epoch": 1.8481182795698925,
-      "grad_norm": 4.430452346801758,
-      "learning_rate": 2.8777777777777776e-05,
-      "loss": 1.6466,
-      "step": 5500
-    },
-    {
-      "epoch": 1.881720430107527,
-      "eval_loss": 1.7603241205215454,
-      "eval_runtime": 190.0315,
-      "eval_samples_per_second": 36.952,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.9727208705935503,
-      "step": 5600
-    },
-    {
-      "epoch": 1.9489247311827957,
-      "eval_loss": 1.8200371265411377,
-      "eval_runtime": 190.0189,
-      "eval_samples_per_second": 36.954,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.9943661223242403,
-      "step": 5800
-    },
-    {
-      "epoch": 2.0161290322580645,
-      "grad_norm": 8.303234100341797,
-      "learning_rate": 1.0555555555555555e-06,
-      "loss": 1.8511,
-      "step": 6000
-    },
-    {
-      "epoch": 2.0161290322580645,
-      "eval_loss": 1.8412970304489136,
-      "eval_runtime": 189.9189,
-      "eval_samples_per_second": 36.974,
-      "eval_steps_per_second": 0.579,
-      "eval_wer": 0.9956417172696953,
-      "step": 6000
-    },
-    {
-      "epoch": 2.0161290322580645,
-      "step": 6000,
-      "total_flos": 1.3379687773166643e+19,
-      "train_loss": 1.3496905517578126,
-      "train_runtime": 9965.7858,
-      "train_samples_per_second": 9.633,
-      "train_steps_per_second": 0.602
+      "epoch": 1.0080645161290323,
+      "step": 3000,
+      "total_flos": 6.709869296482936e+18,
+      "train_loss": 1.1787635701497396,
+      "train_runtime": 4443.9077,
+      "train_samples_per_second": 10.801,
+      "train_steps_per_second": 0.675
     }
   ],
   "logging_steps": 500,
-  "max_steps": 6000,
+  "max_steps": 3000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
+  "num_train_epochs": 2,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -389,7 +212,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.3379687773166643e+19,
+  "total_flos": 6.709869296482936e+18,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null