xlsr53-ptbr-mupe-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: 1.4832
- Wer: 0.5878
- Cer: 0.3261
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 |
|---|---|---|---|---|---|
| 1.6438 | 1.0 | 1000 | 1.5940 | 0.6624 | 0.3650 |
| 1.4311 | 2.0 | 2000 | 1.5149 | 0.6379 | 0.3488 |
| 1.29 | 3.0 | 3000 | 1.4733 | 0.6117 | 0.3317 |
| 1.1968 | 4.0 | 4000 | 1.4522 | 0.6163 | 0.3330 |
| 0.8005 | 5.0 | 5000 | 1.4504 | 0.5965 | 0.3297 |
| 0.5489 | 6.0 | 6000 | 1.4832 | 0.5878 | 0.3260 |
| 0.8353 | 7.0 | 7000 | 1.4658 | 0.5936 | 0.3248 |
| 1.7251 | 8.0 | 8000 | 1.4922 | 0.5901 | 0.3261 |
| 1.0679 | 9.0 | 9000 | 1.4867 | 0.5913 | 0.3238 |
Framework versions
- Transformers 4.55.0
- Pytorch 2.7.0+cu126
- Datasets 4.0.0
- Tokenizers 0.21.4