Whisper Small zh-TW - Fine-tune-nan
This model is a fine-tuned version of openai/whisper-small on the Common Voice 15, 16 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3416
- Cer: 25.1639
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 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: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.3958 | 0.4660 | 1000 | 0.4066 | 30.3417 |
| 0.2337 | 0.9320 | 2000 | 0.3511 | 26.6925 |
| 0.1156 | 1.3979 | 3000 | 0.3464 | 26.7451 |
| 0.0799 | 1.8639 | 4000 | 0.3416 | 25.1639 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0