whisper-small-ptbr-mupe-final3
This model is a fine-tuned version of lejonck/whisper-small-ptbr-mupe-final2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.1030
- Wer: 0.3515
- Cer: 0.5754
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 |
|---|---|---|---|---|---|
| 0.4787 | 1.0 | 500 | 0.8731 | 0.3570 | 0.5753 |
| 0.2399 | 2.0 | 1000 | 0.8857 | 0.3621 | 0.5794 |
| 0.0581 | 3.0 | 1500 | 0.9773 | 0.3665 | 0.5786 |
| 0.015 | 4.0 | 2000 | 1.0512 | 0.4353 | 0.5929 |
| 0.0142 | 5.0 | 2500 | 1.0925 | 0.3503 | 0.5749 |
| 0.0033 | 6.0 | 3000 | 1.1178 | 0.3498 | 0.5749 |
| 0.0048 | 7.0 | 3500 | 1.1313 | 0.3503 | 0.5743 |
| 0.0035 | 8.0 | 4000 | 1.1704 | 0.3541 | 0.5761 |
| 0.0005 | 9.0 | 4500 | 1.1901 | 0.3521 | 0.5756 |
| 0.0029 | 10.0 | 5000 | 1.2289 | 0.3561 | 0.5761 |
| 0.0003 | 11.0 | 5500 | 1.2302 | 0.3728 | 0.5795 |
| 0.0003 | 12.0 | 6000 | 1.2441 | 0.3610 | 0.5777 |
Framework versions
- Transformers 4.55.1
- Pytorch 2.7.0+cu126
- Datasets 4.0.0
- Tokenizers 0.21.4