Whisper Base Hu
This model is a fine-tuned version of openai/whisper-base on the Common Voice 16.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1658
- Wer Ortho: 15.5426
- Wer: 13.6822
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: 2.75e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 0.4422 | 0.33 | 1000 | 0.4764 | 45.9412 | 42.4400 |
| 0.3173 | 0.67 | 2000 | 0.3564 | 36.3754 | 33.3205 |
| 0.2388 | 1.0 | 3000 | 0.2880 | 30.5883 | 27.4711 |
| 0.1896 | 1.34 | 4000 | 0.2556 | 26.9210 | 24.5153 |
| 0.1854 | 1.67 | 5000 | 0.2250 | 24.1720 | 21.5031 |
| 0.0934 | 2.01 | 6000 | 0.1930 | 20.7003 | 18.1026 |
| 0.0885 | 2.34 | 7000 | 0.1875 | 18.9721 | 16.5105 |
| 0.0959 | 2.68 | 8000 | 0.1766 | 17.9002 | 15.7249 |
| 0.0466 | 3.01 | 9000 | 0.1658 | 15.8256 | 13.9134 |
| 0.0484 | 3.35 | 10000 | 0.1658 | 15.5426 | 13.6822 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.0+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0
- Downloads last month
- 0