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-normalized
 - generated_from_trainer
 metrics:
 - wer
@@ -15,10 +17,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # wav2vec2-xlsr-53-ft-btb-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-NORMALIZED - DEFAULT dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.9104
-- Wer: 0.8611
+- Loss: 0.6298
+- Wer: 0.4414
 
 ## Model description
 

all_results.json

@@ -1,14 +1,15 @@
 {
-    "epoch": 5.66,
-    "eval_loss": 0.5991869568824768,
-    "eval_runtime": 213.631,
+    "epoch": 5.657708628005658,
+    "eval_loss": 0.6297643780708313,
+    "eval_runtime": 170.703,
     "eval_samples": 5656,
-    "eval_samples_per_second": 26.476,
-    "eval_steps_per_second": 0.829,
-    "eval_wer": 0.44248649327805,
-    "train_loss": 1.8628226623535156,
-    "train_runtime": 6582.3574,
+    "eval_samples_per_second": 33.134,
+    "eval_steps_per_second": 1.037,
+    "eval_wer": 0.4413827414100239,
+    "total_flos": 1.5333464503060175e+19,
+    "train_loss": 2.1353194427490236,
+    "train_runtime": 5639.522,
     "train_samples": 22621,
-    "train_samples_per_second": 48.615,
-    "train_steps_per_second": 0.76
+    "train_samples_per_second": 56.742,
+    "train_steps_per_second": 0.887
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 5.66,
-    "eval_loss": 0.5991869568824768,
-    "eval_runtime": 213.631,
+    "epoch": 5.657708628005658,
+    "eval_loss": 0.6297643780708313,
+    "eval_runtime": 170.703,
     "eval_samples": 5656,
-    "eval_samples_per_second": 26.476,
-    "eval_steps_per_second": 0.829,
-    "eval_wer": 0.44248649327805
+    "eval_samples_per_second": 33.134,
+    "eval_steps_per_second": 1.037,
+    "eval_wer": 0.4413827414100239
 }

runs/May10_07-39-17_6b3e2401f443/events.out.tfevents.1715329035.6b3e2401f443.30.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b278532640d54823aa4be70dd191911fe1c0b2762f913f5e618a8de06c8b1758
+size 406

train_results.json

@@ -1,8 +1,9 @@
 {
-    "epoch": 5.66,
-    "train_loss": 1.8628226623535156,
-    "train_runtime": 6582.3574,
+    "epoch": 5.657708628005658,
+    "total_flos": 1.5333464503060175e+19,
+    "train_loss": 2.1353194427490236,
+    "train_runtime": 5639.522,
     "train_samples": 22621,
-    "train_samples_per_second": 48.615,
-    "train_steps_per_second": 0.76
+    "train_samples_per_second": 56.742,
+    "train_steps_per_second": 0.887
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.5991869568824768,
+  "best_metric": 0.6297643780708313,
   "best_model_checkpoint": "/models/hfhub/DewiBrynJones/wav2vec2-xlsr-53-ft-btb-cy/checkpoint-1000",
   "epoch": 5.657708628005658,
   "eval_steps": 500,
@@ -9,86 +9,93 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.13,
-      "learning_rate": 0.00014812499999999998,
-      "loss": 4.6574,
+      "epoch": 1.1315417256011315,
+      "grad_norm": 1.0976253747940063,
+      "learning_rate": 0.000147375,
+      "loss": 4.9658,
       "step": 400
     },
     {
-      "epoch": 1.41,
-      "eval_loss": 1.0889060497283936,
-      "eval_runtime": 233.5812,
-      "eval_samples_per_second": 24.214,
-      "eval_steps_per_second": 0.758,
-      "eval_wer": 0.7293001633371027,
+      "epoch": 1.4144271570014144,
+      "eval_loss": 1.1673808097839355,
+      "eval_runtime": 174.6298,
+      "eval_samples_per_second": 32.389,
+      "eval_steps_per_second": 1.014,
+      "eval_wer": 0.802891945242413,
       "step": 500
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 0.00029662499999999996,
-      "loss": 1.0887,
+      "epoch": 2.263083451202263,
+      "grad_norm": 2.727048397064209,
+      "learning_rate": 0.00029587499999999997,
+      "loss": 1.2376,
       "step": 800
     },
     {
-      "epoch": 2.83,
-      "eval_loss": 0.5991869568824768,
-      "eval_runtime": 234.4544,
-      "eval_samples_per_second": 24.124,
-      "eval_steps_per_second": 0.755,
-      "eval_wer": 0.44248649327805,
+      "epoch": 2.828854314002829,
+      "eval_loss": 0.6297643780708313,
+      "eval_runtime": 181.3206,
+      "eval_samples_per_second": 31.193,
+      "eval_steps_per_second": 0.976,
+      "eval_wer": 0.4413827414100239,
       "step": 1000
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 0.0002722142857142857,
-      "loss": 0.7223,
+      "epoch": 3.3946251768033946,
+      "grad_norm": 4.206020832061768,
+      "learning_rate": 0.0002723571428571428,
+      "loss": 0.8262,
       "step": 1200
     },
     {
-      "epoch": 4.24,
-      "eval_loss": 0.726735532283783,
-      "eval_runtime": 213.9298,
-      "eval_samples_per_second": 26.439,
-      "eval_steps_per_second": 0.827,
-      "eval_wer": 0.46312350797838925,
+      "epoch": 4.243281471004243,
+      "eval_loss": 1.9106061458587646,
+      "eval_runtime": 170.5081,
+      "eval_samples_per_second": 33.171,
+      "eval_steps_per_second": 1.038,
+      "eval_wer": 0.8612444030748985,
       "step": 1500
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 0.0002439285714285714,
-      "loss": 0.7162,
+      "epoch": 4.526166902404526,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0002462142857142857,
+      "loss": 1.653,
       "step": 1600
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 0.00021564285714285712,
-      "loss": 2.1295,
+      "epoch": 5.657708628005658,
+      "grad_norm": 0.0,
+      "learning_rate": 0.00021792857142857142,
+      "loss": 1.9941,
       "step": 2000
     },
     {
-      "epoch": 5.66,
-      "eval_loss": 3.0491700172424316,
-      "eval_runtime": 217.4025,
-      "eval_samples_per_second": 26.016,
-      "eval_steps_per_second": 0.814,
-      "eval_wer": 0.9983980399547682,
+      "epoch": 5.657708628005658,
+      "eval_loss": 1.9103525876998901,
+      "eval_runtime": 170.4356,
+      "eval_samples_per_second": 33.186,
+      "eval_steps_per_second": 1.039,
+      "eval_wer": 0.8610999662980854,
       "step": 2000
     },
     {
-      "epoch": 5.66,
+      "epoch": 5.657708628005658,
       "step": 2000,
       "total_flos": 1.5333464503060175e+19,
-      "train_loss": 1.8628226623535156,
-      "train_runtime": 6582.3574,
-      "train_samples_per_second": 48.615,
-      "train_steps_per_second": 0.76
+      "train_loss": 2.1353194427490236,
+      "train_runtime": 5639.522,
+      "train_samples_per_second": 56.742,
+      "train_steps_per_second": 0.887
     }
   ],
   "logging_steps": 400,
   "max_steps": 5000,
+  "num_input_tokens_seen": 0,
   "num_train_epochs": 15,
   "save_steps": 500,
   "total_flos": 1.5333464503060175e+19,
+  "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null
 }