whisper-khanacademy-large-v3-turbo-tr
This model is a fine-tuned version of openai/whisper-large-v3-turbo on the ysdede/khanacademy-turkish dataset. It achieves the following results on the evaluation set:
- Loss: 0.2129
- Wer: 15.6951
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-06
- train_batch_size: 64
- eval_batch_size: 32
- 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: 0.15
- training_steps: 1204
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2298 | 0.1429 | 172 | 0.2418 | 16.5877 |
| 0.2157 | 0.2857 | 344 | 0.2255 | 15.9611 |
| 0.1668 | 1.0939 | 516 | 0.2227 | 15.7461 |
| 0.1752 | 1.2367 | 688 | 0.2159 | 15.8846 |
| 0.1492 | 2.0449 | 860 | 0.2187 | 15.7571 |
| 0.1592 | 2.1877 | 1032 | 0.2134 | 15.5421 |
| 0.1336 | 2.3306 | 1204 | 0.2129 | 15.6951 |
Framework versions
- Transformers 4.48.0.dev0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0