openai/whisper-medium
This model is a fine-tuned version of openai/whisper-medium on the pphuc25/EngMed dataset. It achieves the following results on the evaluation set:
- Loss: 1.3174
- Wer: 16.8776
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.5925 | 1.0 | 323 | 0.7699 | 21.1498 |
| 0.3437 | 2.0 | 646 | 0.8143 | 36.4451 |
| 0.1856 | 3.0 | 969 | 0.8772 | 20.0422 |
| 0.123 | 4.0 | 1292 | 0.8963 | 22.3629 |
| 0.0975 | 5.0 | 1615 | 0.9559 | 20.4641 |
| 0.0746 | 6.0 | 1938 | 0.9947 | 20.7806 |
| 0.0594 | 7.0 | 2261 | 1.0422 | 18.5654 |
| 0.0405 | 8.0 | 2584 | 1.1209 | 18.9873 |
| 0.0352 | 9.0 | 2907 | 1.1780 | 23.7869 |
| 0.0313 | 10.0 | 3230 | 1.1915 | 19.5148 |
| 0.0213 | 11.0 | 3553 | 1.2019 | 16.8776 |
| 0.0154 | 12.0 | 3876 | 1.1514 | 17.8797 |
| 0.0047 | 13.0 | 4199 | 1.1908 | 19.6730 |
| 0.0091 | 14.0 | 4522 | 1.1906 | 18.7764 |
| 0.0017 | 15.0 | 4845 | 1.2421 | 17.9852 |
| 0.0012 | 16.0 | 5168 | 1.2742 | 18.0907 |
| 0.0003 | 17.0 | 5491 | 1.3038 | 17.5105 |
| 0.0003 | 18.0 | 5814 | 1.3091 | 17.4578 |
| 0.0002 | 19.0 | 6137 | 1.3155 | 16.9304 |
| 0.0 | 20.0 | 6460 | 1.3174 | 16.8776 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.19.1