Whisper Small Burmese - SLR80 Fine-Tuned
This model is a fine-tuned version of openai/whisper-small on the Myanmar Speech Dataset (OpenSLR-80) dataset. It achieves the following results on the evaluation set:
- Loss: 0.3932
- Wer: 161.2199
- Cer: 90.5602
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: 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: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.8707 | 0.7018 | 100 | 0.8376 | 219.3678 | 91.2284 |
| 0.6616 | 1.4 | 200 | 0.6328 | 258.8157 | 91.5565 |
| 0.4185 | 2.0982 | 300 | 0.3932 | 161.2199 | 90.5602 |
| 0.284 | 2.8 | 400 | 0.2824 | 184.5058 | 90.4603 |
| 0.2265 | 3.4982 | 500 | 0.2524 | 173.4194 | 90.1916 |
Framework versions
- Transformers 4.57.3
- Pytorch 2.9.0+cu126
- Datasets 4.4.2
- Tokenizers 0.22.1