whisper-small-common-voice-1
This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4217
- Wer: 0.3988
- Cer: 0.3678
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 |
|---|---|---|---|---|---|
| 0.1611 | 1.0 | 500 | 0.3533 | 0.4579 | 0.3761 |
| 0.0402 | 2.0 | 1000 | 0.3580 | 0.5923 | 0.3872 |
| 0.0295 | 3.0 | 1500 | 0.3920 | 0.4593 | 0.3727 |
| 0.0168 | 4.0 | 2000 | 0.4245 | 0.4077 | 0.3657 |
| 0.0135 | 5.0 | 2500 | 0.4114 | 0.4177 | 0.3678 |
| 0.0092 | 6.0 | 3000 | 0.4214 | 0.4371 | 0.3680 |
| 0.0187 | 7.0 | 3500 | 0.4234 | 0.4697 | 0.3764 |
| 0.002 | 8.0 | 4000 | 0.4290 | 0.4697 | 0.3839 |
| 0.0005 | 9.0 | 4500 | 0.4217 | 0.3988 | 0.3678 |
| 0.0004 | 10.0 | 5000 | 0.4277 | 0.4163 | 0.3718 |
| 0.0003 | 11.0 | 5500 | 0.4302 | 0.4166 | 0.3721 |
| 0.0003 | 12.0 | 6000 | 0.4314 | 0.4200 | 0.3728 |
Framework versions
- Transformers 4.55.2
- Pytorch 2.7.0+cu126
- Datasets 2.19.1
- Tokenizers 0.21.4