whisper-small-pashto-stage2-fleurs
This model is a fine-tuned version of Adnan666/whisper-small-pashto-run11-cv24only on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5410
- Wer: 36.1158
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- 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 |
|---|---|---|---|---|
| 0.7706 | 0.3175 | 50 | 0.6587 | 38.4127 |
| 0.6458 | 0.6349 | 100 | 0.5812 | 37.5163 |
| 0.6419 | 0.9524 | 150 | 0.5461 | 36.4893 |
| 0.4854 | 1.2698 | 200 | 0.5395 | 36.4332 |
| 0.4769 | 1.5873 | 250 | 0.5286 | 35.8170 |
| 0.4429 | 1.9079 | 300 | 0.5293 | 35.9290 |
| 0.3465 | 2.2254 | 350 | 0.5291 | 35.6489 |
| 0.3558 | 2.5429 | 400 | 0.5352 | 36.2278 |
| 0.3389 | 2.8603 | 450 | 0.5328 | 36.0411 |
| 0.2698 | 3.1778 | 500 | 0.5410 | 36.1158 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu118
- Datasets 2.19.0
- Tokenizers 0.19.1