whisper-small-somali
This model is a fine-tuned version of Zyroxx66/whisper-small-somali on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1304
- Wer: 9.8739
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: 32
- 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: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2316 | 1.2903 | 200 | 0.2340 | 25.6653 |
| 0.0663 | 2.5806 | 400 | 0.1504 | 14.2857 |
| 0.0253 | 3.8710 | 600 | 0.1323 | 11.2745 |
| 0.0088 | 5.1613 | 800 | 0.1244 | 10.0140 |
| 0.0080 | 6.4516 | 1000 | 0.1288 | 10.8894 |
| 0.0052 | 7.7419 | 1200 | 0.1278 | 10.0840 |
| 0.0027 | 9.0323 | 1400 | 0.1300 | 9.9090 |
| 0.0015 | 10.3226 | 1600 | 0.1291 | 9.8739 |
| 0.0022 | 11.6129 | 1800 | 0.1303 | 9.8739 |
| 0.0020 | 12.9032 | 2000 | 0.1304 | 9.8739 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.10.0+cu128
- Datasets 5.0.1
- Tokenizers 0.22.2