whisper-small-arabic-cv_11-20_02
This model is a fine-tuned version of openai/whisper-small on common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.3309
- Wer: 45.2778
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: 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: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.28 | 0.42 | 1000 | 0.3918 | 53.7563 |
| 0.2354 | 0.83 | 2000 | 0.3460 | 47.3767 |
| 0.1736 | 1.25 | 3000 | 0.3309 | 45.2778 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.12.0
- Datasets 2.1.0
- Tokenizers 0.12.1