Whisper Small - Swedish
This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4915
- Wer: 25.5384
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-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 6000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2107 | 1.3 | 1000 | 0.4673 | 34.0432 |
| 0.0821 | 2.59 | 2000 | 0.4284 | 27.4152 |
| 0.0378 | 3.89 | 3000 | 0.4210 | 25.3637 |
| 0.0042 | 5.18 | 4000 | 0.4247 | 23.5541 |
| 0.001 | 6.48 | 5000 | 0.4286 | 22.7770 |
| 0.0106 | 7.77 | 6000 | 0.4915 | 25.5384 |
Framework versions
- Transformers 4.25.0.dev0
- Pytorch 1.12.1
- Datasets 2.7.1
- Tokenizers 0.13.2