Whisper Small - Hanif Rahman
This model is a fine-tuned version of openai/whisper-small on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.8094
- Wer Ortho: 51.6855
- Wer: 47.9806
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: 5e-05
- train_batch_size: 32
- 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_ratio: 0.1
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 0.6754 | 0.9346 | 100 | 0.6689 | 62.1021 | 58.4888 |
| 0.4477 | 1.8692 | 200 | 0.6215 | 57.3134 | 53.5101 |
| 0.2243 | 2.8037 | 300 | 0.6222 | 55.8883 | 52.0928 |
| 0.0949 | 3.7383 | 400 | 0.6822 | 54.6007 | 49.6989 |
| 0.0448 | 4.6729 | 500 | 0.7240 | 53.5301 | 49.4346 |
| 0.0201 | 5.6075 | 600 | 0.7355 | 52.7344 | 48.9646 |
| 0.0124 | 6.5421 | 700 | 0.7615 | 52.3944 | 48.6929 |
| 0.0035 | 7.4766 | 800 | 0.7868 | 51.0778 | 47.2243 |
| 0.002 | 8.4112 | 900 | 0.8025 | 51.6276 | 47.6869 |
| 0.0011 | 9.3458 | 1000 | 0.8094 | 51.6855 | 47.9806 |
Framework versions
- Transformers 4.46.2
- Pytorch 2.5.0+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3