Whisper Small PT with Common Voice 11
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: 0.3487
- Wer: 14.3802
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: 8
- seed: 42
- gradient_accumulation_steps: 2
- 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: 1000
- training_steps: 10000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.1202 | 0.88 | 1000 | 0.2225 | 15.5847 |
| 0.1024 | 1.76 | 2000 | 0.2160 | 15.0651 |
| 0.0832 | 2.64 | 3000 | 0.2259 | 15.0923 |
| 0.0081 | 3.51 | 4000 | 0.2519 | 14.7345 |
| 0.0387 | 4.39 | 5000 | 0.2718 | 14.7311 |
| 0.0039 | 5.27 | 6000 | 0.3031 | 14.5914 |
| 0.001 | 6.15 | 7000 | 0.3238 | 14.5710 |
| 0.0007 | 7.03 | 8000 | 0.3285 | 14.5113 |
| 0.0009 | 7.91 | 9000 | 0.3467 | 14.3580 |
| 0.0008 | 8.79 | 10000 | 0.3487 | 14.3802 |
Framework versions
- Transformers 4.25.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.6.1
- Tokenizers 0.12.1