whisper-medium-medical
This model is a fine-tuned version of openai/whisper-medium on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2889
- Wer: 0.1281
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: 4
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_BNB 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: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.4747 | 1.0 | 102 | 0.3153 | 0.1621 |
| 1.4045 | 2.0 | 204 | 0.2641 | 0.1341 |
| 0.6465 | 3.0 | 306 | 0.2671 | 0.1321 |
| 0.2759 | 4.0 | 408 | 0.2727 | 0.1278 |
| 0.1574 | 5.0 | 510 | 0.2889 | 0.1281 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 2.19.0
- Tokenizers 0.22.2