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: 0.5226
 - Wer: 0.4001

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 0.38476337052712584,
-    "eval_loss": 0.5865370631217957,
-    "eval_runtime": 179.3337,
+    "epoch": 0.6412722842118763,
+    "eval_loss": 0.5226185917854309,
+    "eval_runtime": 188.9199,
     "eval_samples": 7022,
-    "eval_samples_per_second": 39.156,
-    "eval_steps_per_second": 2.448,
-    "eval_wer": 0.44523578575319894,
-    "total_flos": 6.835515215409967e+18,
-    "train_loss": 0.9274428431193034,
-    "train_runtime": 4544.4955,
+    "eval_samples_per_second": 37.169,
+    "eval_steps_per_second": 0.582,
+    "eval_wer": 0.40009832711037885,
+    "total_flos": 1.1393778193380235e+19,
+    "train_loss": 0.7283544036865235,
+    "train_runtime": 7737.7643,
     "train_samples": 124748,
-    "train_samples_per_second": 10.562,
-    "train_steps_per_second": 1.32
+    "train_samples_per_second": 10.339,
+    "train_steps_per_second": 1.292
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.38476337052712584,
-    "eval_loss": 0.5865370631217957,
-    "eval_runtime": 179.3337,
+    "epoch": 0.6412722842118763,
+    "eval_loss": 0.5226185917854309,
+    "eval_runtime": 188.9199,
     "eval_samples": 7022,
-    "eval_samples_per_second": 39.156,
-    "eval_steps_per_second": 2.448,
-    "eval_wer": 0.44523578575319894
+    "eval_samples_per_second": 37.169,
+    "eval_steps_per_second": 0.582,
+    "eval_wer": 0.40009832711037885
 }

runs/Aug30_21-53-10_4b35055fdbcb/events.out.tfevents.1725061483.4b35055fdbcb.318.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bce9a3f5d305870094b0b15d0e6f2c1a88995a9bae5068167ab796c85fe77df5
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.38476337052712584,
-    "total_flos": 6.835515215409967e+18,
-    "train_loss": 0.9274428431193034,
-    "train_runtime": 4544.4955,
+    "epoch": 0.6412722842118763,
+    "total_flos": 1.1393778193380235e+19,
+    "train_loss": 0.7283544036865235,
+    "train_runtime": 7737.7643,
     "train_samples": 124748,
-    "train_samples_per_second": 10.562,
-    "train_steps_per_second": 1.32
+    "train_samples_per_second": 10.339,
+    "train_steps_per_second": 1.292
 }

trainer_state.json

@@ -1,217 +1,345 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.38476337052712584,
+  "epoch": 0.6412722842118763,
   "eval_steps": 500,
-  "global_step": 6000,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.03206361421059382,
-      "grad_norm": 10.316516876220703,
+      "grad_norm": 11.533838272094727,
       "learning_rate": 0.0002465,
-      "loss": 4.7051,
+      "loss": 4.6618,
       "step": 500
     },
     {
       "epoch": 0.03206361421059382,
-      "eval_loss": 1.6561740636825562,
-      "eval_runtime": 191.0416,
-      "eval_samples_per_second": 36.756,
-      "eval_steps_per_second": 2.298,
-      "eval_wer": 0.9354894431230816,
+      "eval_loss": 1.5996026992797852,
+      "eval_runtime": 185.1067,
+      "eval_samples_per_second": 37.935,
+      "eval_steps_per_second": 0.594,
+      "eval_wer": 0.9161163448889834,
       "step": 500
     },
     {
       "epoch": 0.06412722842118763,
-      "grad_norm": 4.912766933441162,
-      "learning_rate": 0.00027816666666666663,
-      "loss": 1.0362,
+      "grad_norm": 4.801280975341797,
+      "learning_rate": 0.0002874574468085106,
+      "loss": 1.0278,
       "step": 1000
     },
     {
       "epoch": 0.06412722842118763,
-      "eval_loss": 1.1839842796325684,
-      "eval_runtime": 175.5788,
-      "eval_samples_per_second": 39.993,
-      "eval_steps_per_second": 2.5,
-      "eval_wer": 0.7977517639086356,
+      "eval_loss": 1.1463252305984497,
+      "eval_runtime": 184.8935,
+      "eval_samples_per_second": 37.979,
+      "eval_steps_per_second": 0.595,
+      "eval_wer": 0.7792157748574922,
       "step": 1000
     },
     {
       "epoch": 0.09619084263178146,
-      "grad_norm": 4.178534030914307,
-      "learning_rate": 0.00025038888888888886,
-      "loss": 0.811,
+      "grad_norm": 4.746730327606201,
+      "learning_rate": 0.0002715,
+      "loss": 0.8164,
       "step": 1500
     },
     {
       "epoch": 0.09619084263178146,
-      "eval_loss": 1.0081175565719604,
-      "eval_runtime": 175.5534,
-      "eval_samples_per_second": 39.999,
-      "eval_steps_per_second": 2.501,
-      "eval_wer": 0.7174690070290597,
+      "eval_loss": 1.0590689182281494,
+      "eval_runtime": 185.3865,
+      "eval_samples_per_second": 37.878,
+      "eval_steps_per_second": 0.593,
+      "eval_wer": 0.7363238948165668,
       "step": 1500
     },
     {
       "epoch": 0.12825445684237527,
-      "grad_norm": 7.957609176635742,
-      "learning_rate": 0.0002226111111111111,
-      "loss": 0.6903,
+      "grad_norm": 21.847057342529297,
+      "learning_rate": 0.00025554255319148935,
+      "loss": 0.7124,
       "step": 2000
     },
     {
       "epoch": 0.12825445684237527,
-      "eval_loss": 0.8935067057609558,
-      "eval_runtime": 176.1028,
-      "eval_samples_per_second": 39.874,
-      "eval_steps_per_second": 2.493,
-      "eval_wer": 0.6401094885661516,
+      "eval_loss": 0.9373884797096252,
+      "eval_runtime": 186.5741,
+      "eval_samples_per_second": 37.637,
+      "eval_steps_per_second": 0.59,
+      "eval_wer": 0.6622596632960842,
       "step": 2000
     },
     {
       "epoch": 0.16031807105296908,
-      "grad_norm": 6.541042327880859,
-      "learning_rate": 0.00019483333333333332,
-      "loss": 0.6238,
+      "grad_norm": 4.238761901855469,
+      "learning_rate": 0.0002395851063829787,
+      "loss": 0.6566,
       "step": 2500
     },
     {
       "epoch": 0.16031807105296908,
-      "eval_loss": 0.8059775233268738,
-      "eval_runtime": 177.0587,
-      "eval_samples_per_second": 39.659,
-      "eval_steps_per_second": 2.479,
-      "eval_wer": 0.5848868573858276,
+      "eval_loss": 0.8721805810928345,
+      "eval_runtime": 187.0392,
+      "eval_samples_per_second": 37.543,
+      "eval_steps_per_second": 0.588,
+      "eval_wer": 0.615168949893036,
       "step": 2500
     },
     {
       "epoch": 0.19238168526356292,
-      "grad_norm": 5.036514759063721,
-      "learning_rate": 0.00016705555555555554,
-      "loss": 0.5649,
+      "grad_norm": 5.65878963470459,
+      "learning_rate": 0.00022362765957446805,
+      "loss": 0.6101,
       "step": 3000
     },
     {
       "epoch": 0.19238168526356292,
-      "eval_loss": 0.7770201563835144,
-      "eval_runtime": 178.1454,
-      "eval_samples_per_second": 39.417,
-      "eval_steps_per_second": 2.464,
-      "eval_wer": 0.5589231852668783,
+      "eval_loss": 0.8194323182106018,
+      "eval_runtime": 187.1888,
+      "eval_samples_per_second": 37.513,
+      "eval_steps_per_second": 0.588,
+      "eval_wer": 0.5927264513214366,
       "step": 3000
     },
     {
       "epoch": 0.22444529947415673,
-      "grad_norm": 4.140881538391113,
-      "learning_rate": 0.00013927777777777777,
-      "loss": 0.5309,
+      "grad_norm": 6.294471740722656,
+      "learning_rate": 0.00020767021276595744,
+      "loss": 0.5777,
       "step": 3500
     },
     {
       "epoch": 0.22444529947415673,
-      "eval_loss": 0.7264481782913208,
-      "eval_runtime": 176.7301,
-      "eval_samples_per_second": 39.733,
-      "eval_steps_per_second": 2.484,
-      "eval_wer": 0.5326539018589139,
+      "eval_loss": 0.7799355387687683,
+      "eval_runtime": 187.6349,
+      "eval_samples_per_second": 37.424,
+      "eval_steps_per_second": 0.586,
+      "eval_wer": 0.5707490134070343,
       "step": 3500
     },
     {
       "epoch": 0.25650891368475054,
-      "grad_norm": 6.205782413482666,
-      "learning_rate": 0.00011155555555555555,
-      "loss": 0.4892,
+      "grad_norm": 3.848400115966797,
+      "learning_rate": 0.0001917446808510638,
+      "loss": 0.5431,
       "step": 4000
     },
     {
       "epoch": 0.25650891368475054,
-      "eval_loss": 0.6865007877349854,
-      "eval_runtime": 178.7785,
-      "eval_samples_per_second": 39.278,
-      "eval_steps_per_second": 2.456,
-      "eval_wer": 0.5106366016024662,
+      "eval_loss": 0.7453157901763916,
+      "eval_runtime": 187.5991,
+      "eval_samples_per_second": 37.431,
+      "eval_steps_per_second": 0.586,
+      "eval_wer": 0.550233194700966,
       "step": 4000
     },
     {
       "epoch": 0.2885725278953444,
-      "grad_norm": 9.618443489074707,
-      "learning_rate": 8.388888888888888e-05,
-      "loss": 0.4521,
+      "grad_norm": 9.198270797729492,
+      "learning_rate": 0.00017585106382978722,
+      "loss": 0.512,
       "step": 4500
     },
     {
       "epoch": 0.2885725278953444,
-      "eval_loss": 0.6477864980697632,
-      "eval_runtime": 177.7072,
-      "eval_samples_per_second": 39.514,
-      "eval_steps_per_second": 2.47,
-      "eval_wer": 0.4861345486918508,
+      "eval_loss": 0.7209311127662659,
+      "eval_runtime": 188.2502,
+      "eval_samples_per_second": 37.301,
+      "eval_steps_per_second": 0.584,
+      "eval_wer": 0.5346735938558843,
       "step": 4500
     },
     {
       "epoch": 0.32063614210593816,
-      "grad_norm": 8.743291854858398,
-      "learning_rate": 5.61111111111111e-05,
-      "loss": 0.4309,
+      "grad_norm": 7.854990482330322,
+      "learning_rate": 0.00015989361702127658,
+      "loss": 0.4953,
       "step": 5000
     },
     {
       "epoch": 0.32063614210593816,
-      "eval_loss": 0.6221807599067688,
-      "eval_runtime": 177.9115,
-      "eval_samples_per_second": 39.469,
-      "eval_steps_per_second": 2.468,
-      "eval_wer": 0.4763416999960138,
+      "eval_loss": 0.6942155957221985,
+      "eval_runtime": 187.3133,
+      "eval_samples_per_second": 37.488,
+      "eval_steps_per_second": 0.587,
+      "eval_wer": 0.5225022920846676,
       "step": 5000
     },
     {
       "epoch": 0.352699756316532,
-      "grad_norm": 8.870036125183105,
-      "learning_rate": 2.833333333333333e-05,
-      "loss": 0.4055,
+      "grad_norm": 10.654298782348633,
+      "learning_rate": 0.00014393617021276595,
+      "loss": 0.4746,
       "step": 5500
     },
     {
       "epoch": 0.352699756316532,
-      "eval_loss": 0.5988152027130127,
-      "eval_runtime": 178.6515,
-      "eval_samples_per_second": 39.306,
-      "eval_steps_per_second": 2.457,
-      "eval_wer": 0.45257045668956536,
+      "eval_loss": 0.6680377721786499,
+      "eval_runtime": 188.6546,
+      "eval_samples_per_second": 37.221,
+      "eval_steps_per_second": 0.583,
+      "eval_wer": 0.495714798230112,
       "step": 5500
     },
     {
       "epoch": 0.38476337052712584,
-      "grad_norm": 14.018965721130371,
-      "learning_rate": 6.11111111111111e-07,
-      "loss": 0.3896,
+      "grad_norm": 19.936861038208008,
+      "learning_rate": 0.00012801063829787234,
+      "loss": 0.4535,
       "step": 6000
     },
     {
       "epoch": 0.38476337052712584,
-      "eval_loss": 0.5865370631217957,
-      "eval_runtime": 176.8507,
-      "eval_samples_per_second": 39.706,
-      "eval_steps_per_second": 2.482,
-      "eval_wer": 0.44523578575319894,
+      "eval_loss": 0.6434958577156067,
+      "eval_runtime": 188.0277,
+      "eval_samples_per_second": 37.346,
+      "eval_steps_per_second": 0.585,
+      "eval_wer": 0.47308627539563375,
       "step": 6000
     },
     {
-      "epoch": 0.38476337052712584,
-      "step": 6000,
-      "total_flos": 6.835515215409967e+18,
-      "train_loss": 0.9274428431193034,
-      "train_runtime": 4544.4955,
-      "train_samples_per_second": 10.562,
-      "train_steps_per_second": 1.32
+      "epoch": 0.4168269847377196,
+      "grad_norm": 6.950649261474609,
+      "learning_rate": 0.0001120531914893617,
+      "loss": 0.4249,
+      "step": 6500
+    },
+    {
+      "epoch": 0.4168269847377196,
+      "eval_loss": 0.6293027400970459,
+      "eval_runtime": 187.8772,
+      "eval_samples_per_second": 37.375,
+      "eval_steps_per_second": 0.585,
+      "eval_wer": 0.4816832538301067,
+      "step": 6500
+    },
+    {
+      "epoch": 0.44889059894831346,
+      "grad_norm": 5.574142932891846,
+      "learning_rate": 9.612765957446806e-05,
+      "loss": 0.4065,
+      "step": 7000
+    },
+    {
+      "epoch": 0.44889059894831346,
+      "eval_loss": 0.5997486710548401,
+      "eval_runtime": 188.7633,
+      "eval_samples_per_second": 37.2,
+      "eval_steps_per_second": 0.583,
+      "eval_wer": 0.44947448145736724,
+      "step": 7000
+    },
+    {
+      "epoch": 0.4809542131589073,
+      "grad_norm": 5.602737903594971,
+      "learning_rate": 8.017021276595744e-05,
+      "loss": 0.393,
+      "step": 7500
+    },
+    {
+      "epoch": 0.4809542131589073,
+      "eval_loss": 0.5802670121192932,
+      "eval_runtime": 189.0166,
+      "eval_samples_per_second": 37.15,
+      "eval_steps_per_second": 0.582,
+      "eval_wer": 0.44260487117819797,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5130178273695011,
+      "grad_norm": 15.019088745117188,
+      "learning_rate": 6.424468085106383e-05,
+      "loss": 0.3808,
+      "step": 8000
+    },
+    {
+      "epoch": 0.5130178273695011,
+      "eval_loss": 0.5661880970001221,
+      "eval_runtime": 189.1322,
+      "eval_samples_per_second": 37.127,
+      "eval_steps_per_second": 0.582,
+      "eval_wer": 0.4299153589603901,
+      "step": 8000
+    },
+    {
+      "epoch": 0.5450814415800949,
+      "grad_norm": 3.79109525680542,
+      "learning_rate": 4.8287234042553194e-05,
+      "loss": 0.3722,
+      "step": 8500
+    },
+    {
+      "epoch": 0.5450814415800949,
+      "eval_loss": 0.553141176700592,
+      "eval_runtime": 188.8506,
+      "eval_samples_per_second": 37.183,
+      "eval_steps_per_second": 0.582,
+      "eval_wer": 0.4194182755550831,
+      "step": 8500
+    },
+    {
+      "epoch": 0.5771450557906888,
+      "grad_norm": 12.16059398651123,
+      "learning_rate": 3.232978723404255e-05,
+      "loss": 0.3622,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5771450557906888,
+      "eval_loss": 0.5399113297462463,
+      "eval_runtime": 188.6804,
+      "eval_samples_per_second": 37.216,
+      "eval_steps_per_second": 0.583,
+      "eval_wer": 0.40727354867856336,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6092086700012825,
+      "grad_norm": 7.863190650939941,
+      "learning_rate": 1.6372340425531912e-05,
+      "loss": 0.3526,
+      "step": 9500
+    },
+    {
+      "epoch": 0.6092086700012825,
+      "eval_loss": 0.5277913808822632,
+      "eval_runtime": 188.1946,
+      "eval_samples_per_second": 37.312,
+      "eval_steps_per_second": 0.585,
+      "eval_wer": 0.40279567892212226,
+      "step": 9500
+    },
+    {
+      "epoch": 0.6412722842118763,
+      "grad_norm": 5.124056339263916,
+      "learning_rate": 4.1489361702127654e-07,
+      "loss": 0.3337,
+      "step": 10000
+    },
+    {
+      "epoch": 0.6412722842118763,
+      "eval_loss": 0.5226185917854309,
+      "eval_runtime": 188.4518,
+      "eval_samples_per_second": 37.262,
+      "eval_steps_per_second": 0.584,
+      "eval_wer": 0.40009832711037885,
+      "step": 10000
+    },
+    {
+      "epoch": 0.6412722842118763,
+      "step": 10000,
+      "total_flos": 1.1393778193380235e+19,
+      "train_loss": 0.7283544036865235,
+      "train_runtime": 7737.7643,
+      "train_samples_per_second": 10.339,
+      "train_steps_per_second": 1.292
     }
   ],
   "logging_steps": 500,
-  "max_steps": 6000,
+  "max_steps": 10000,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 500,
@@ -227,7 +355,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6.835515215409967e+18,
+  "total_flos": 1.1393778193380235e+19,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null