whisper-small-pashto
This model is a fine-tuned version of ihanif/whisper_small_ps_augmented on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.3364
- Wer: 32.6050
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: 5e-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 600
- training_steps: 21854
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.321 | 0.0137 | 100 | 0.2794 | 28.2726 |
| 1.1509 | 0.0275 | 200 | 0.3045 | 29.4678 |
| 1.0585 | 0.0412 | 300 | 0.3154 | 30.4388 |
| 1.0012 | 0.0549 | 400 | 0.3218 | 30.4575 |
| 0.9466 | 0.0686 | 500 | 0.3329 | 32.4370 |
| 0.9275 | 0.0824 | 600 | 0.3364 | 32.6050 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu118
- Datasets 2.19.0
- Tokenizers 0.19.1