WhisperTiny15hCommonVoice7.5hAugment
This model is a fine-tuned version of OpenAI/whisper-tiny on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.8838
- Wer: 0.4254
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: 16
- eval_batch_size: 8
- 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: 500
- training_steps: 5000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8948 | 0.5952 | 500 | 1.0324 | 0.4926 |
| 0.643 | 1.1905 | 1000 | 0.9414 | 0.4496 |
| 0.5849 | 1.7857 | 1500 | 0.9019 | 0.4379 |
| 0.4744 | 2.3810 | 2000 | 0.9003 | 0.4375 |
| 0.458 | 2.9762 | 2500 | 0.8847 | 0.4391 |
| 0.3704 | 3.5714 | 3000 | 0.8810 | 0.4272 |
| 0.3121 | 4.1667 | 3500 | 0.8789 | 0.4256 |
| 0.2938 | 4.7619 | 4000 | 0.8812 | 0.4241 |
| 0.2695 | 5.3571 | 4500 | 0.8847 | 0.4280 |
| 0.2766 | 5.9524 | 5000 | 0.8838 | 0.4254 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.6.0
- Datasets 3.3.0
- Tokenizers 0.21.0