whisper-small-multilingual-25h
This model is a fine-tuned version of openai/whisper-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7039
- Wer: 80.9142
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_FUSED 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 | Wer |
|---|---|---|---|---|
| 1.1817 | 0.3455 | 500 | 1.1716 | 102.1129 |
| 0.9381 | 0.6911 | 1000 | 0.9055 | 87.1129 |
| 0.6638 | 1.0366 | 1500 | 0.8039 | 75.0140 |
| 0.6351 | 1.3822 | 2000 | 0.7635 | 82.1642 |
| 0.6050 | 1.7277 | 2500 | 0.7280 | 74.7108 |
| 0.4512 | 2.0733 | 3000 | 0.7156 | 83.0597 |
| 0.4346 | 2.4188 | 3500 | 0.7098 | 81.8937 |
| 0.4388 | 2.7643 | 4000 | 0.7039 | 80.9142 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 4.8.3
- Tokenizers 0.22.2