Whisper Base Shona + Tshivenda + isiZulu Pilot
This model is a fine-tuned version of openai/whisper-base on the Cleaned WAXAL Shona + Swivuriso Tshivenda and isiZulu dataset. It achieves the following results on the evaluation set:
- Loss: 0.5226
- Wer: 46.0310
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
- optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8060 | 0.3008 | 1000 | 0.7400 | 58.4445 |
| 0.6281 | 0.6015 | 2000 | 0.6038 | 50.9763 |
| 0.6247 | 0.9023 | 3000 | 0.5537 | 46.8376 |
| 0.4988 | 1.2030 | 4000 | 0.5315 | 45.7078 |
| 0.4905 | 1.5038 | 5000 | 0.5226 | 46.0310 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.13.0+cu130
- Datasets 5.0.1
- Tokenizers 0.22.2