Model save

README.md

@@ -2,8 +2,6 @@
 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
@@ -17,10 +15,10 @@ 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 the DEWIBRYNJONES/BANC-TRAWSGRIFIADAU-BANGOR-CLEAN-WITH-CCV - DEFAULT dataset.
+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.
 It achieves the following results on the evaluation set:
-- Loss: 0.5865
-- Wer: 0.4452
+- Loss: 0.5226
+- Wer: 0.4001
 
 ## Model description
 
@@ -41,30 +39,38 @@ More information needed
 The following hyperparameters were used during training:
 - learning_rate: 0.0003
 - train_batch_size: 8
-- eval_batch_size: 16
+- eval_batch_size: 64
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 600
-- training_steps: 6000
+- training_steps: 10000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch  | Step | Validation Loss | Wer    |
-|:-------------:|:------:|:----:|:---------------:|:------:|
-| 4.7051        | 0.0321 | 500  | 1.6562          | 0.9355 |
-| 1.0362        | 0.0641 | 1000 | 1.1840          | 0.7978 |
-| 0.811         | 0.0962 | 1500 | 1.0081          | 0.7175 |
-| 0.6903        | 0.1283 | 2000 | 0.8935          | 0.6401 |
-| 0.6238        | 0.1603 | 2500 | 0.8060          | 0.5849 |
-| 0.5649        | 0.1924 | 3000 | 0.7770          | 0.5589 |
-| 0.5309        | 0.2244 | 3500 | 0.7264          | 0.5327 |
-| 0.4892        | 0.2565 | 4000 | 0.6865          | 0.5106 |
-| 0.4521        | 0.2886 | 4500 | 0.6478          | 0.4861 |
-| 0.4309        | 0.3206 | 5000 | 0.6222          | 0.4763 |
-| 0.4055        | 0.3527 | 5500 | 0.5988          | 0.4526 |
-| 0.3896        | 0.3848 | 6000 | 0.5865          | 0.4452 |
+| Training Loss | Epoch  | Step  | Validation Loss | Wer    |
+|:-------------:|:------:|:-----:|:---------------:|:------:|
+| 4.6618        | 0.0321 | 500   | 1.5996          | 0.9161 |
+| 1.0278        | 0.0641 | 1000  | 1.1463          | 0.7792 |
+| 0.8164        | 0.0962 | 1500  | 1.0591          | 0.7363 |
+| 0.7124        | 0.1283 | 2000  | 0.9374          | 0.6623 |
+| 0.6566        | 0.1603 | 2500  | 0.8722          | 0.6152 |
+| 0.6101        | 0.1924 | 3000  | 0.8194          | 0.5927 |
+| 0.5777        | 0.2244 | 3500  | 0.7799          | 0.5707 |
+| 0.5431        | 0.2565 | 4000  | 0.7453          | 0.5502 |
+| 0.512         | 0.2886 | 4500  | 0.7209          | 0.5347 |
+| 0.4953        | 0.3206 | 5000  | 0.6942          | 0.5225 |
+| 0.4746        | 0.3527 | 5500  | 0.6680          | 0.4957 |
+| 0.4535        | 0.3848 | 6000  | 0.6435          | 0.4731 |
+| 0.4249        | 0.4168 | 6500  | 0.6293          | 0.4817 |
+| 0.4065        | 0.4489 | 7000  | 0.5997          | 0.4495 |
+| 0.393         | 0.4810 | 7500  | 0.5803          | 0.4426 |
+| 0.3808        | 0.5130 | 8000  | 0.5662          | 0.4299 |
+| 0.3722        | 0.5451 | 8500  | 0.5531          | 0.4194 |
+| 0.3622        | 0.5771 | 9000  | 0.5399          | 0.4073 |
+| 0.3526        | 0.6092 | 9500  | 0.5278          | 0.4028 |
+| 0.3337        | 0.6413 | 10000 | 0.5226          | 0.4001 |
 
 
 ### Framework versions