whisper-small-pashto-stage2-v2
This model is a fine-tuned version of Adnan666/whisper-small-pashto-run11-epoch2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5200
- Wer: 35.9104
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: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH 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
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.1448 | 0.6329 | 50 | 0.5839 | 38.0019 |
| 2.3491 | 1.2658 | 100 | 0.5228 | 35.8730 |
| 2.0525 | 1.8987 | 150 | 0.5060 | 35.5369 |
| 1.5750 | 2.5316 | 200 | 0.5048 | 35.1821 |
| 1.3983 | 3.1646 | 250 | 0.5097 | 35.6489 |
| 1.4097 | 3.7975 | 300 | 0.5155 | 36.3585 |
| 1.3779 | 4.4304 | 350 | 0.5200 | 35.9104 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.6.0+cu124
- Datasets 4.8.5
- Tokenizers 0.22.2