WhisperTiny15hCommonVoice15hAugment
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: 0.8652
- Wer: 0.4183
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 |
|---|---|---|---|---|
| 0.9492 | 0.4464 | 500 | 1.0368 | 0.4824 |
| 0.7759 | 0.8929 | 1000 | 0.9355 | 0.4505 |
| 0.6086 | 1.3393 | 1500 | 0.8993 | 0.4374 |
| 0.5636 | 1.7857 | 2000 | 0.8855 | 0.4282 |
| 0.4343 | 2.2321 | 2500 | 0.8726 | 0.4342 |
| 0.4228 | 2.6786 | 3000 | 0.8714 | 0.4204 |
| 0.3585 | 3.125 | 3500 | 0.8591 | 0.4194 |
| 0.3302 | 3.5714 | 4000 | 0.8674 | 0.4179 |
| 0.3235 | 4.0179 | 4500 | 0.8640 | 0.4186 |
| 0.3032 | 4.4643 | 5000 | 0.8652 | 0.4183 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.6.0
- Datasets 3.3.0
- Tokenizers 0.21.0