whisper-small-ptbr-cv-final2
This model is a fine-tuned version of lejonck/whisper-small-ptbr-cv-final1 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5542
- Wer: 0.3751
- Cer: 0.5790
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: 8
- 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 |
|---|---|---|---|---|---|
| 0.351 | 1.0 | 375 | 0.3592 | 0.3804 | 0.5755 |
| 0.0842 | 2.0 | 750 | 0.3901 | 0.3857 | 0.5761 |
| 0.0366 | 3.0 | 1125 | 0.4186 | 0.3768 | 0.5775 |
| 0.0186 | 4.0 | 1500 | 0.4657 | 0.3880 | 0.5842 |
| 0.0109 | 5.0 | 1875 | 0.5108 | 0.3810 | 0.5788 |
| 0.0026 | 6.0 | 2250 | 0.5285 | 0.3827 | 0.5823 |
| 0.0009 | 7.0 | 2625 | 0.5498 | 0.3792 | 0.5786 |
| 0.0007 | 8.0 | 3000 | 0.5474 | 0.3733 | 0.5777 |
| 0.0006 | 9.0 | 3375 | 0.5659 | 0.3739 | 0.5793 |
| 0.0005 | 10.0 | 3750 | 0.5588 | 0.3792 | 0.5812 |
| 0.0005 | 11.0 | 4125 | 0.5657 | 0.3786 | 0.5794 |
| 0.0005 | 12.0 | 4500 | 0.5700 | 0.3768 | 0.5794 |
Framework versions
- Transformers 4.55.0
- Pytorch 2.7.0+cu126
- Datasets 2.19.1
- Tokenizers 0.21.4