whisper-small-E50_pause_speed
This model is a fine-tuned version of openai/whisper-small on the aihub old adult freq speed pause changed dataset. It achieves the following results on the evaluation set:
- Loss: 0.1714
- Cer: 8.0886
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: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- 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: 50
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.3029 | 0.1289 | 100 | 0.2462 | 6.1971 |
| 0.161 | 0.2579 | 200 | 0.2166 | 5.6861 |
| 0.1967 | 0.3868 | 300 | 0.2087 | 5.7625 |
| 0.1561 | 0.5158 | 400 | 0.2014 | 5.3513 |
| 0.1372 | 0.6447 | 500 | 0.1973 | 5.5686 |
| 0.1191 | 0.7737 | 600 | 0.1859 | 5.2455 |
| 0.1175 | 0.9026 | 700 | 0.1801 | 5.3865 |
| 0.0697 | 1.0309 | 800 | 0.1754 | 5.1281 |
| 0.0558 | 1.1599 | 900 | 0.1793 | 4.8578 |
| 0.0571 | 1.2888 | 1000 | 0.1746 | 4.8931 |
| 0.0542 | 1.4178 | 1100 | 0.1765 | 5.7096 |
| 0.0514 | 1.5467 | 1200 | 0.1744 | 7.9300 |
| 0.0527 | 1.6757 | 1300 | 0.1751 | 9.3867 |
| 0.0469 | 1.8046 | 1400 | 0.1712 | 12.9171 |
| 0.0478 | 1.9336 | 1500 | 0.1714 | 8.0886 |
Framework versions
- Transformers 4.47.0.dev0
- Pytorch 2.5.0+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3