End of training

README.md

@@ -2,6 +2,8 @@
 license: mit
 base_model: facebook/w2v-bert-2.0
 tags:
+- automatic-speech-recognition
+- DewiBrynJones/banc-trawsgrifiadau-bangor-normalized
 - generated_from_trainer
 metrics:
 - wer
@@ -15,7 +17,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # w2v2-bert-ft-btb-cy
 
-This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on an unknown dataset.
+This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the DEWIBRYNJONES/BANC-TRAWSGRIFIADAU-BANGOR-NORMALIZED - DEFAULT dataset.
 It achieves the following results on the evaluation set:
 - Loss: 2.9177
 - Wer: 1.0

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 3.536067892503536,
-    "eval_loss": 2.9374959468841553,
-    "eval_runtime": 134.4964,
+    "epoch": 10.0,
+    "eval_loss": 2.91774320602417,
+    "eval_runtime": 135.1853,
     "eval_samples": 5656,
-    "eval_samples_per_second": 42.053,
-    "eval_steps_per_second": 5.257,
+    "eval_samples_per_second": 41.839,
+    "eval_steps_per_second": 5.23,
     "eval_wer": 1.0,
-    "total_flos": 8.753895488690287e+18,
-    "train_loss": 3.32313603515625,
-    "train_runtime": 6689.3091,
+    "total_flos": 2.4662883830172946e+19,
+    "train_loss": 3.228043903960534,
+    "train_runtime": 12366.081,
     "train_samples": 22621,
-    "train_samples_per_second": 11.959,
-    "train_steps_per_second": 0.374
+    "train_samples_per_second": 18.293,
+    "train_steps_per_second": 0.572
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.536067892503536,
-    "eval_loss": 2.9374959468841553,
-    "eval_runtime": 134.4964,
+    "epoch": 10.0,
+    "eval_loss": 2.91774320602417,
+    "eval_runtime": 135.1853,
     "eval_samples": 5656,
-    "eval_samples_per_second": 42.053,
-    "eval_steps_per_second": 5.257,
+    "eval_samples_per_second": 41.839,
+    "eval_steps_per_second": 5.23,
     "eval_wer": 1.0
 }

runs/May13_06-46-35_09e070d6a7b1/events.out.tfevents.1715592220.09e070d6a7b1.1206.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d46ade80e1997b90c18e4024f5ed2f998ec93d5e19212ec846553cb3ae02755
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.536067892503536,
-    "total_flos": 8.753895488690287e+18,
-    "train_loss": 3.32313603515625,
-    "train_runtime": 6689.3091,
+    "epoch": 10.0,
+    "total_flos": 2.4662883830172946e+19,
+    "train_loss": 3.228043903960534,
+    "train_runtime": 12366.081,
     "train_samples": 22621,
-    "train_samples_per_second": 11.959,
-    "train_steps_per_second": 0.374
+    "train_samples_per_second": 18.293,
+    "train_steps_per_second": 0.572
 }

trainer_state.json

@@ -1,289 +1,334 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.536067892503536,
-  "eval_steps": 100,
-  "global_step": 2500,
+  "epoch": 10.0,
+  "eval_steps": 300,
+  "global_step": 7070,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.14144271570014144,
-      "eval_loss": 6.970242977142334,
-      "eval_runtime": 137.617,
-      "eval_samples_per_second": 41.1,
-      "eval_steps_per_second": 5.137,
+      "epoch": 0.4243281471004243,
+      "eval_loss": 5.990268707275391,
+      "eval_runtime": 138.1214,
+      "eval_samples_per_second": 40.949,
+      "eval_steps_per_second": 5.119,
       "eval_wer": 1.0,
-      "step": 100
+      "step": 300
     },
     {
-      "epoch": 0.2828854314002829,
-      "eval_loss": 3.36755633354187,
-      "eval_runtime": 133.5904,
-      "eval_samples_per_second": 42.338,
-      "eval_steps_per_second": 5.292,
-      "eval_wer": 1.0,
-      "step": 200
+      "epoch": 0.7072135785007072,
+      "grad_norm": 2.022184371948242,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 7.061,
+      "step": 500
     },
     {
-      "epoch": 0.4243281471004243,
-      "eval_loss": 3.0392866134643555,
-      "eval_runtime": 133.8525,
-      "eval_samples_per_second": 42.255,
-      "eval_steps_per_second": 5.282,
+      "epoch": 0.8486562942008486,
+      "eval_loss": 3.0451483726501465,
+      "eval_runtime": 135.2034,
+      "eval_samples_per_second": 41.833,
+      "eval_steps_per_second": 5.229,
       "eval_wer": 1.0,
-      "step": 300
+      "step": 600
     },
     {
-      "epoch": 0.5657708628005658,
-      "eval_loss": 2.9887070655822754,
-      "eval_runtime": 134.092,
-      "eval_samples_per_second": 42.18,
-      "eval_steps_per_second": 5.272,
+      "epoch": 1.272984441301273,
+      "eval_loss": 2.9642043113708496,
+      "eval_runtime": 136.8192,
+      "eval_samples_per_second": 41.339,
+      "eval_steps_per_second": 5.167,
       "eval_wer": 1.0,
-      "step": 400
+      "step": 900
     },
     {
-      "epoch": 0.7072135785007072,
-      "grad_norm": 0.6687237620353699,
-      "learning_rate": 0.00029699999999999996,
-      "loss": 4.717,
-      "step": 500
+      "epoch": 1.4144271570014144,
+      "grad_norm": 0.7561541795730591,
+      "learning_rate": 4.624048706240488e-05,
+      "loss": 3.0081,
+      "step": 1000
     },
     {
-      "epoch": 0.7072135785007072,
-      "eval_loss": 3.0226564407348633,
-      "eval_runtime": 134.7626,
-      "eval_samples_per_second": 41.97,
-      "eval_steps_per_second": 5.246,
+      "epoch": 1.6973125884016973,
+      "eval_loss": 2.956415891647339,
+      "eval_runtime": 136.6446,
+      "eval_samples_per_second": 41.392,
+      "eval_steps_per_second": 5.174,
       "eval_wer": 1.0,
-      "step": 500
+      "step": 1200
     },
     {
-      "epoch": 0.8486562942008486,
-      "eval_loss": 3.0405795574188232,
-      "eval_runtime": 135.9502,
-      "eval_samples_per_second": 41.603,
-      "eval_steps_per_second": 5.2,
+      "epoch": 2.1216407355021216,
+      "grad_norm": 0.3668934106826782,
+      "learning_rate": 4.245053272450533e-05,
+      "loss": 2.9733,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1216407355021216,
+      "eval_loss": 2.947998285293579,
+      "eval_runtime": 136.4831,
+      "eval_samples_per_second": 41.441,
+      "eval_steps_per_second": 5.18,
       "eval_wer": 1.0,
-      "step": 600
+      "step": 1500
     },
     {
-      "epoch": 0.9900990099009901,
-      "eval_loss": 3.0028798580169678,
-      "eval_runtime": 135.0397,
-      "eval_samples_per_second": 41.884,
-      "eval_steps_per_second": 5.235,
+      "epoch": 2.545968882602546,
+      "eval_loss": 2.945077657699585,
+      "eval_runtime": 136.3573,
+      "eval_samples_per_second": 41.479,
+      "eval_steps_per_second": 5.185,
       "eval_wer": 1.0,
-      "step": 700
+      "step": 1800
+    },
+    {
+      "epoch": 2.828854314002829,
+      "grad_norm": 1.441468596458435,
+      "learning_rate": 3.866057838660579e-05,
+      "loss": 2.9454,
+      "step": 2000
     },
     {
-      "epoch": 1.1315417256011315,
-      "eval_loss": 2.948317050933838,
-      "eval_runtime": 135.3684,
-      "eval_samples_per_second": 41.782,
-      "eval_steps_per_second": 5.223,
+      "epoch": 2.9702970297029703,
+      "eval_loss": 2.9147346019744873,
+      "eval_runtime": 135.861,
+      "eval_samples_per_second": 41.631,
+      "eval_steps_per_second": 5.204,
       "eval_wer": 1.0,
-      "step": 800
+      "step": 2100
     },
     {
-      "epoch": 1.272984441301273,
-      "eval_loss": 2.9510738849639893,
-      "eval_runtime": 135.3897,
-      "eval_samples_per_second": 41.776,
-      "eval_steps_per_second": 5.222,
+      "epoch": 3.3946251768033946,
+      "eval_loss": 2.9019417762756348,
+      "eval_runtime": 136.0133,
+      "eval_samples_per_second": 41.584,
+      "eval_steps_per_second": 5.198,
       "eval_wer": 1.0,
-      "step": 900
+      "step": 2400
     },
     {
-      "epoch": 1.4144271570014144,
-      "grad_norm": 0.3827725648880005,
-      "learning_rate": 0.00022574999999999996,
-      "loss": 3.0065,
-      "step": 1000
+      "epoch": 3.536067892503536,
+      "grad_norm": 0.46694883704185486,
+      "learning_rate": 3.487062404870624e-05,
+      "loss": 2.9064,
+      "step": 2500
     },
     {
-      "epoch": 1.4144271570014144,
-      "eval_loss": 2.9472572803497314,
-      "eval_runtime": 134.8251,
-      "eval_samples_per_second": 41.951,
-      "eval_steps_per_second": 5.244,
+      "epoch": 3.818953323903819,
+      "eval_loss": 2.884958505630493,
+      "eval_runtime": 136.0891,
+      "eval_samples_per_second": 41.561,
+      "eval_steps_per_second": 5.195,
       "eval_wer": 1.0,
-      "step": 1000
+      "step": 2700
     },
     {
-      "epoch": 1.5558698727015559,
-      "eval_loss": 2.9448139667510986,
-      "eval_runtime": 135.1998,
-      "eval_samples_per_second": 41.834,
-      "eval_steps_per_second": 5.229,
+      "epoch": 4.243281471004243,
+      "grad_norm": 0.4952280819416046,
+      "learning_rate": 3.10882800608828e-05,
+      "loss": 2.9048,
+      "step": 3000
+    },
+    {
+      "epoch": 4.243281471004243,
+      "eval_loss": 2.8812334537506104,
+      "eval_runtime": 136.4568,
+      "eval_samples_per_second": 41.449,
+      "eval_steps_per_second": 5.181,
       "eval_wer": 1.0,
-      "step": 1100
+      "step": 3000
     },
     {
-      "epoch": 1.6973125884016973,
-      "eval_loss": 2.946981191635132,
-      "eval_runtime": 135.1335,
-      "eval_samples_per_second": 41.855,
-      "eval_steps_per_second": 5.232,
+      "epoch": 4.667609618104668,
+      "eval_loss": 2.884371042251587,
+      "eval_runtime": 136.787,
+      "eval_samples_per_second": 41.349,
+      "eval_steps_per_second": 5.169,
       "eval_wer": 1.0,
-      "step": 1200
+      "step": 3300
+    },
+    {
+      "epoch": 4.9504950495049505,
+      "grad_norm": 0.8865047097206116,
+      "learning_rate": 2.7290715372907157e-05,
+      "loss": 2.8965,
+      "step": 3500
     },
     {
-      "epoch": 1.8387553041018387,
-      "eval_loss": 2.944624662399292,
-      "eval_runtime": 135.5596,
-      "eval_samples_per_second": 41.723,
-      "eval_steps_per_second": 5.215,
+      "epoch": 5.091937765205092,
+      "eval_loss": 2.9125277996063232,
+      "eval_runtime": 136.3564,
+      "eval_samples_per_second": 41.48,
+      "eval_steps_per_second": 5.185,
       "eval_wer": 1.0,
-      "step": 1300
+      "step": 3600
     },
     {
-      "epoch": 1.9801980198019802,
-      "eval_loss": 2.9431614875793457,
-      "eval_runtime": 135.6031,
-      "eval_samples_per_second": 41.71,
-      "eval_steps_per_second": 5.214,
+      "epoch": 5.516265912305516,
+      "eval_loss": 2.898144006729126,
+      "eval_runtime": 136.0768,
+      "eval_samples_per_second": 41.565,
+      "eval_steps_per_second": 5.196,
       "eval_wer": 1.0,
-      "step": 1400
+      "step": 3900
     },
     {
-      "epoch": 2.1216407355021216,
-      "grad_norm": 0.5749518275260925,
-      "learning_rate": 0.0001512,
-      "loss": 2.9634,
-      "step": 1500
+      "epoch": 5.657708628005658,
+      "grad_norm": 0.3529145121574402,
+      "learning_rate": 2.3508371385083716e-05,
+      "loss": 2.9261,
+      "step": 4000
     },
     {
-      "epoch": 2.1216407355021216,
-      "eval_loss": 2.9475975036621094,
-      "eval_runtime": 136.0526,
-      "eval_samples_per_second": 41.572,
-      "eval_steps_per_second": 5.197,
+      "epoch": 5.9405940594059405,
+      "eval_loss": 2.905318260192871,
+      "eval_runtime": 136.6781,
+      "eval_samples_per_second": 41.382,
+      "eval_steps_per_second": 5.173,
       "eval_wer": 1.0,
-      "step": 1500
+      "step": 4200
     },
     {
-      "epoch": 2.263083451202263,
-      "eval_loss": 2.9624321460723877,
-      "eval_runtime": 136.0863,
-      "eval_samples_per_second": 41.562,
-      "eval_steps_per_second": 5.195,
+      "epoch": 6.364922206506365,
+      "grad_norm": 0.22229251265525818,
+      "learning_rate": 1.971841704718417e-05,
+      "loss": 2.9273,
+      "step": 4500
+    },
+    {
+      "epoch": 6.364922206506365,
+      "eval_loss": 2.916677951812744,
+      "eval_runtime": 136.7502,
+      "eval_samples_per_second": 41.36,
+      "eval_steps_per_second": 5.17,
       "eval_wer": 1.0,
-      "step": 1600
+      "step": 4500
     },
     {
-      "epoch": 2.4045261669024045,
-      "eval_loss": 2.9580626487731934,
-      "eval_runtime": 136.566,
-      "eval_samples_per_second": 41.416,
-      "eval_steps_per_second": 5.177,
+      "epoch": 6.789250353606789,
+      "eval_loss": 2.911259651184082,
+      "eval_runtime": 136.484,
+      "eval_samples_per_second": 41.441,
+      "eval_steps_per_second": 5.18,
       "eval_wer": 1.0,
-      "step": 1700
+      "step": 4800
     },
     {
-      "epoch": 2.545968882602546,
-      "eval_loss": 2.9552838802337646,
-      "eval_runtime": 136.3048,
-      "eval_samples_per_second": 41.495,
-      "eval_steps_per_second": 5.187,
+      "epoch": 7.072135785007072,
+      "grad_norm": 1.7586228847503662,
+      "learning_rate": 1.592846270928463e-05,
+      "loss": 2.9302,
+      "step": 5000
+    },
+    {
+      "epoch": 7.2135785007072135,
+      "eval_loss": 2.9133317470550537,
+      "eval_runtime": 135.9523,
+      "eval_samples_per_second": 41.603,
+      "eval_steps_per_second": 5.2,
       "eval_wer": 1.0,
-      "step": 1800
+      "step": 5100
     },
     {
-      "epoch": 2.6874115983026874,
-      "eval_loss": 2.9515163898468018,
-      "eval_runtime": 136.0496,
-      "eval_samples_per_second": 41.573,
-      "eval_steps_per_second": 5.197,
+      "epoch": 7.637906647807638,
+      "eval_loss": 2.921302080154419,
+      "eval_runtime": 136.5228,
+      "eval_samples_per_second": 41.429,
+      "eval_steps_per_second": 5.179,
       "eval_wer": 1.0,
-      "step": 1900
+      "step": 5400
     },
     {
-      "epoch": 2.828854314002829,
-      "grad_norm": 0.206673726439476,
-      "learning_rate": 7.664999999999999e-05,
-      "loss": 2.9677,
-      "step": 2000
+      "epoch": 7.779349363507779,
+      "grad_norm": 0.6302638649940491,
+      "learning_rate": 1.2146118721461187e-05,
+      "loss": 2.9397,
+      "step": 5500
     },
     {
-      "epoch": 2.828854314002829,
-      "eval_loss": 2.9480981826782227,
-      "eval_runtime": 134.8264,
-      "eval_samples_per_second": 41.95,
-      "eval_steps_per_second": 5.244,
+      "epoch": 8.062234794908063,
+      "eval_loss": 2.9251174926757812,
+      "eval_runtime": 136.4335,
+      "eval_samples_per_second": 41.456,
+      "eval_steps_per_second": 5.182,
       "eval_wer": 1.0,
-      "step": 2000
+      "step": 5700
     },
     {
-      "epoch": 2.9702970297029703,
-      "eval_loss": 2.9509432315826416,
-      "eval_runtime": 134.773,
-      "eval_samples_per_second": 41.967,
-      "eval_steps_per_second": 5.246,
-      "eval_wer": 1.0,
-      "step": 2100
+      "epoch": 8.486562942008486,
+      "grad_norm": 0.5835816860198975,
+      "learning_rate": 8.340943683409437e-06,
+      "loss": 2.937,
+      "step": 6000
     },
     {
-      "epoch": 3.1117397454031117,
-      "eval_loss": 2.940824031829834,
-      "eval_runtime": 135.008,
-      "eval_samples_per_second": 41.894,
-      "eval_steps_per_second": 5.237,
+      "epoch": 8.486562942008486,
+      "eval_loss": 2.921030282974243,
+      "eval_runtime": 136.1229,
+      "eval_samples_per_second": 41.551,
+      "eval_steps_per_second": 5.194,
       "eval_wer": 1.0,
-      "step": 2200
+      "step": 6000
     },
     {
-      "epoch": 3.253182461103253,
-      "eval_loss": 2.9393398761749268,
-      "eval_runtime": 134.9605,
-      "eval_samples_per_second": 41.909,
-      "eval_steps_per_second": 5.239,
+      "epoch": 8.910891089108912,
+      "eval_loss": 2.92145037651062,
+      "eval_runtime": 137.1094,
+      "eval_samples_per_second": 41.252,
+      "eval_steps_per_second": 5.156,
       "eval_wer": 1.0,
-      "step": 2300
+      "step": 6300
     },
     {
-      "epoch": 3.3946251768033946,
-      "eval_loss": 2.938136339187622,
-      "eval_runtime": 134.6374,
-      "eval_samples_per_second": 42.009,
-      "eval_steps_per_second": 5.251,
-      "eval_wer": 1.0,
-      "step": 2400
+      "epoch": 9.193776520509195,
+      "grad_norm": 0.7211419939994812,
+      "learning_rate": 4.558599695585997e-06,
+      "loss": 2.9406,
+      "step": 6500
     },
     {
-      "epoch": 3.536067892503536,
-      "grad_norm": 0.6203744411468506,
-      "learning_rate": 1.9499999999999995e-06,
-      "loss": 2.9612,
-      "step": 2500
+      "epoch": 9.335219236209335,
+      "eval_loss": 2.917142629623413,
+      "eval_runtime": 136.4229,
+      "eval_samples_per_second": 41.459,
+      "eval_steps_per_second": 5.182,
+      "eval_wer": 1.0,
+      "step": 6600
     },
     {
-      "epoch": 3.536067892503536,
-      "eval_loss": 2.9374959468841553,
-      "eval_runtime": 134.9253,
-      "eval_samples_per_second": 41.919,
-      "eval_steps_per_second": 5.24,
+      "epoch": 9.75954738330976,
+      "eval_loss": 2.9176828861236572,
+      "eval_runtime": 137.2177,
+      "eval_samples_per_second": 41.219,
+      "eval_steps_per_second": 5.152,
       "eval_wer": 1.0,
-      "step": 2500
+      "step": 6900
     },
     {
-      "epoch": 3.536067892503536,
-      "step": 2500,
-      "total_flos": 8.753895488690287e+18,
-      "train_loss": 3.32313603515625,
-      "train_runtime": 6689.3091,
-      "train_samples_per_second": 11.959,
-      "train_steps_per_second": 0.374
+      "epoch": 9.900990099009901,
+      "grad_norm": 0.0,
+      "learning_rate": 7.762557077625571e-07,
+      "loss": 2.9378,
+      "step": 7000
+    },
+    {
+      "epoch": 10.0,
+      "step": 7070,
+      "total_flos": 2.4662883830172946e+19,
+      "train_loss": 3.228043903960534,
+      "train_runtime": 12366.081,
+      "train_samples_per_second": 18.293,
+      "train_steps_per_second": 0.572
     }
   ],
   "logging_steps": 500,
-  "max_steps": 2500,
+  "max_steps": 7070,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
-  "save_steps": 400,
-  "total_flos": 8.753895488690287e+18,
+  "num_train_epochs": 10,
+  "save_steps": 600,
+  "total_flos": 2.4662883830172946e+19,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null