Whisper Small Fine-tuned with Uyghur Common Voice
This model is a fine-tuned version of openai/whisper-small on the Uyghur Common Voice dataset.
As a proof-of-concept, only 3264 recordings (~5.5 hrs of audio) were used for training, and 937 recordings (~1.5 hrs of audio) were used for validation. You may find the full dataset for Uyghur and other languages here: https://commonvoice.mozilla.org/en/datasets.
This model achieves the following results on the evaluation set:
- Loss: 0.5105
- Wer Ortho: 41.6377
- Wer: 34.9961
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: 0.0001
- 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: 50
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 0.0574 | 2.4510 | 500 | 0.5105 | 41.6377 | 34.9961 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.5.0+cu121
- Datasets 3.1.0
- Tokenizers 0.19.1