EYEDOL/whisper-tiny-yoruba3
This model is a fine-tuned version of EYEDOL/whisper-tiny-yoruba2 on the EYEDOL/naija-voices-yoruba-split_0-6 dataset. It achieves the following results on the evaluation set:
- Loss: 0.7111
- Wer Ortho: 0.7096
- Wer: 0.6293
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: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 1.4417 | 1.0 | 583 | 0.7132 | 0.7172 | 0.6352 |
| 1.3179 | 2.0 | 1166 | 0.7039 | 0.6964 | 0.6110 |
| 1.2151 | 3.0 | 1749 | 0.7038 | 0.7101 | 0.6275 |
| 1.1268 | 4.0 | 2332 | 0.7058 | 0.7037 | 0.6166 |
| 1.0504 | 5.0 | 2915 | 0.7111 | 0.7096 | 0.6293 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 4.8.3
- Tokenizers 0.22.2