openai/whisper-tiny
This model is a fine-tuned version of openai/whisper-tiny on the pphuc25/FrenchMed dataset. It achieves the following results on the evaluation set:
- Loss: 1.9773
- Wer: 57.9179
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 |
|---|---|---|---|---|
| 1.4539 | 1.0 | 215 | 1.4469 | 192.2287 |
| 0.9269 | 2.0 | 430 | 1.4258 | 127.1994 |
| 0.5316 | 3.0 | 645 | 1.5228 | 66.6422 |
| 0.3295 | 4.0 | 860 | 1.6796 | 60.4839 |
| 0.212 | 5.0 | 1075 | 1.7495 | 73.7537 |
| 0.1228 | 6.0 | 1290 | 1.8167 | 78.9589 |
| 0.0766 | 7.0 | 1505 | 1.8370 | 80.0587 |
| 0.0617 | 8.0 | 1720 | 1.8817 | 61.1437 |
| 0.0575 | 9.0 | 1935 | 1.9629 | 88.0499 |
| 0.0319 | 10.0 | 2150 | 1.9228 | 58.3578 |
| 0.0266 | 11.0 | 2365 | 1.9362 | 57.1848 |
| 0.0143 | 12.0 | 2580 | 1.9740 | 57.1848 |
| 0.0124 | 13.0 | 2795 | 1.9917 | 86.9501 |
| 0.0109 | 14.0 | 3010 | 1.9632 | 56.5982 |
| 0.0087 | 15.0 | 3225 | 1.9501 | 60.8504 |
| 0.0048 | 16.0 | 3440 | 1.9785 | 55.7918 |
| 0.0034 | 17.0 | 3655 | 1.9765 | 58.6510 |
| 0.0021 | 18.0 | 3870 | 1.9765 | 56.8915 |
| 0.0007 | 19.0 | 4085 | 1.9737 | 58.3578 |
| 0.0007 | 20.0 | 4300 | 1.9773 | 57.9179 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.19.1