<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-xlsr-53-ft-cy-ccv-btb-2306
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the PYTHON/CUSTOM_COMMON_VOICE.PY,TECHIAITH/BANC-TRAWSGRIFIADAU-BANGOR - CY dataset. It achieves the following results on the evaluation set:
- Loss: 0.2105
- Wer: 0.1713
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 800
- training_steps: 5000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.3473 | 0.27 | 400 | 1.3366 | 0.8356 |
| 0.6668 | 0.54 | 800 | 0.4918 | 0.4319 |
| 0.4521 | 0.81 | 1200 | 0.3612 | 0.3137 |
| 0.3764 | 1.08 | 1600 | 0.3145 | 0.2692 |
| 0.3092 | 1.35 | 2000 | 0.2806 | 0.2445 |
| 0.2919 | 1.62 | 2400 | 0.2706 | 0.2334 |
| 0.2743 | 1.89 | 2800 | 0.2423 | 0.2130 |
| 0.2386 | 2.16 | 3200 | 0.2338 | 0.2009 |
| 0.2055 | 2.43 | 3600 | 0.2221 | 0.1903 |
| 0.1982 | 2.7 | 4000 | 0.2188 | 0.1821 |
| 0.1931 | 2.97 | 4400 | 0.2114 | 0.1775 |
| 0.1725 | 3.24 | 4800 | 0.2125 | 0.1733 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.0
- Tokenizers 0.13.3
