Whisper Small dk
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: 0.6207
- Wer: 29.4944
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.4214 | 1.37 | 200 | 0.5155 | 32.5933 |
| 0.1758 | 2.75 | 400 | 0.4674 | 29.5260 |
| 0.0591 | 4.12 | 600 | 0.5032 | 30.5361 |
| 0.0258 | 5.5 | 800 | 0.5336 | 30.0573 |
| 0.017 | 6.87 | 1000 | 0.5676 | 29.2419 |
| 0.0067 | 8.25 | 1200 | 0.5738 | 29.1209 |
| 0.0046 | 9.62 | 1400 | 0.5981 | 29.2839 |
| 0.0027 | 11.0 | 1600 | 0.6114 | 29.4418 |
| 0.0021 | 12.37 | 1800 | 0.6184 | 29.4155 |
| 0.002 | 13.75 | 2000 | 0.6207 | 29.4944 |
Framework versions
- Transformers 4.39.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2