openai/whisper-base
This is an automatic speech recognition model that also does punctuation and casing. This model is for research only, we do not recommend using this model on production environments. See our learnings when training these models.
This model is a fine-tuned version of openai/whisper-base on the mozilla-foundation/common_voice_11_0 ca dataset. It achieves the following results on the evaluation set:
- Loss: 0.3608
- Wer: 16.1510
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: 2
- eval_batch_size: 1
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 40000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.4841 | 0.1 | 4000 | 0.5078 | 26.7974 |
| 0.3116 | 0.2 | 8000 | 0.4524 | 22.9455 |
| 0.3971 | 0.3 | 12000 | 0.4281 | 21.5427 |
| 0.2965 | 0.4 | 16000 | 0.4037 | 20.3082 |
| 0.2634 | 1.09 | 20000 | 0.3875 | 18.7980 |
| 0.2163 | 1.19 | 24000 | 0.3754 | 17.8170 |
| 0.3182 | 1.29 | 28000 | 0.3695 | 16.8587 |
| 0.2201 | 1.39 | 32000 | 0.3613 | 16.5785 |
| 0.155 | 2.08 | 36000 | 0.3633 | 16.3959 |
| 0.0904 | 2.18 | 40000 | 0.3608 | 16.1510 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.10.0+cu102
- Datasets 2.8.0
- Tokenizers 0.13.2