xlsr53-ptbr-cv-final1
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53-portuguese on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4651
- Wer: 0.3284
- Cer: 0.1632
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: 1e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 12
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.4242 | 1.0 | 1000 | 0.6077 | 0.4240 | 0.2238 |
| 0.5916 | 2.0 | 2000 | 0.5118 | 0.3840 | 0.1893 |
| 0.5889 | 3.0 | 3000 | 0.4797 | 0.3529 | 0.1835 |
| 0.3731 | 4.0 | 4000 | 0.4642 | 0.3464 | 0.1722 |
| 0.3221 | 5.0 | 5000 | 0.4652 | 0.3489 | 0.1851 |
| 0.2849 | 6.0 | 6000 | 0.4661 | 0.3431 | 0.1650 |
| 0.2214 | 7.0 | 7000 | 0.4539 | 0.3342 | 0.1695 |
| 0.2706 | 8.0 | 8000 | 0.4584 | 0.3366 | 0.1687 |
| 0.1478 | 9.0 | 9000 | 0.4572 | 0.3333 | 0.1664 |
| 0.1415 | 10.0 | 10000 | 0.4635 | 0.3301 | 0.1655 |
| 0.2655 | 11.0 | 11000 | 0.4651 | 0.3284 | 0.1632 |
| 0.1047 | 12.0 | 12000 | 0.4658 | 0.3301 | 0.1624 |
Framework versions
- Transformers 4.55.0
- Pytorch 2.7.0+cu126
- Datasets 2.19.1
- Tokenizers 0.21.4