whisper-small-mupe-1
This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.9038
- Wer: 0.3488
- Cer: 0.5846
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: 2
- seed: 42
- optimizer: Use OptimizerNames.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_steps: 100
- num_epochs: 12
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.4507 | 1.0 | 500 | 0.7181 | 0.3314 | 0.5808 |
| 0.2228 | 2.0 | 1000 | 0.7651 | 0.3602 | 0.5947 |
| 0.0408 | 3.0 | 1500 | 0.8454 | 0.3473 | 0.5875 |
| 0.0258 | 4.0 | 2000 | 0.8905 | 0.3284 | 0.5842 |
| 0.009 | 5.0 | 2500 | 0.8925 | 0.3359 | 0.6071 |
| 0.0097 | 6.0 | 3000 | 0.9411 | 0.3296 | 0.5859 |
| 0.002 | 7.0 | 3500 | 0.9892 | 0.3302 | 0.5882 |
Framework versions
- Transformers 4.55.2
- Pytorch 2.7.0+cu126
- Datasets 4.0.0
- Tokenizers 0.21.4