openai/whisper-base
This model is a fine-tuned version of openai/whisper-base on the pphuc25/ChiMed dataset. It achieves the following results on the evaluation set:
- Loss: 1.3748
- Cer: 30.8155
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 | Cer |
|---|---|---|---|---|
| 0.8187 | 1.0 | 161 | 1.0315 | 58.4225 |
| 0.4746 | 2.0 | 322 | 1.0785 | 42.6025 |
| 0.29 | 3.0 | 483 | 1.1087 | 44.8752 |
| 0.1474 | 4.0 | 644 | 1.2218 | 34.2914 |
| 0.0994 | 5.0 | 805 | 1.2465 | 55.9269 |
| 0.0779 | 6.0 | 966 | 1.3060 | 36.7201 |
| 0.0549 | 7.0 | 1127 | 1.3262 | 32.8877 |
| 0.0504 | 8.0 | 1288 | 1.3147 | 38.2353 |
| 0.032 | 9.0 | 1449 | 1.4272 | 34.0686 |
| 0.0283 | 10.0 | 1610 | 1.3575 | 31.7513 |
| 0.0244 | 11.0 | 1771 | 1.3715 | 33.2442 |
| 0.0131 | 12.0 | 1932 | 1.3905 | 32.4198 |
| 0.0111 | 13.0 | 2093 | 1.3886 | 33.7344 |
| 0.0072 | 14.0 | 2254 | 1.3618 | 36.1854 |
| 0.0015 | 15.0 | 2415 | 1.3707 | 31.2166 |
| 0.0027 | 16.0 | 2576 | 1.3733 | 30.6373 |
| 0.0007 | 17.0 | 2737 | 1.3673 | 31.3503 |
| 0.0004 | 18.0 | 2898 | 1.3741 | 30.8824 |
| 0.0003 | 19.0 | 3059 | 1.3739 | 30.7709 |
| 0.0003 | 20.0 | 3220 | 1.3748 | 30.8155 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.19.1