real-voices-400-10-epochs
This model is a fine-tuned version of openai/whisper-small on the 400-real-voices dataset. It achieves the following results on the evaluation set:
- Loss: 0.2847
- Wer: 13.9764
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: 4
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0 | 0 | 0.6890 | 41.0325 |
| 0.3321 | 1.0 | 50 | 0.2364 | 16.4233 |
| 0.0869 | 2.0 | 100 | 0.2197 | 14.5017 |
| 0.0354 | 3.0 | 150 | 0.2506 | 14.8604 |
| 0.0176 | 4.0 | 200 | 0.2449 | 14.0405 |
| 0.0097 | 5.0 | 250 | 0.2616 | 14.1430 |
| 0.0054 | 6.0 | 300 | 0.2709 | 14.0277 |
| 0.0027 | 7.0 | 350 | 0.2771 | 13.8611 |
| 0.0023 | 8.0 | 400 | 0.2836 | 13.9380 |
| 0.0019 | 9.0 | 450 | 0.2840 | 13.9252 |
| 0.0017 | 10.0 | 500 | 0.2847 | 13.9764 |
Framework versions
- Transformers 4.49.0
- Pytorch 2.5.1+cu121
- Datasets 3.4.1
- Tokenizers 0.21.0