metadata
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-tiny-ar-quran
results: []
whisper-tiny-ar-quran
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0928
- Wer: 7.0535
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: 0.0001
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
0.1766 | 0.05 | 500 | 0.2829 | 20.0236 |
0.1129 | 0.09 | 1000 | 0.1981 | 13.8364 |
0.0775 | 0.14 | 1500 | 0.1763 | 12.5450 |
0.0678 | 0.19 | 2000 | 0.1485 | 10.7302 |
0.0437 | 0.23 | 2500 | 0.1336 | 9.6693 |
0.0341 | 0.28 | 3000 | 0.1244 | 8.9602 |
0.0302 | 0.33 | 3500 | 0.1059 | 8.2224 |
0.0189 | 0.37 | 4000 | 0.1044 | 7.6902 |
0.0167 | 0.42 | 4500 | 0.0966 | 7.2643 |
0.0151 | 0.47 | 5000 | 0.0928 | 7.0535 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2