wav2vec2-base-arabic-finetuned-continued-v3
This model is a fine-tuned version of Mohammadawad1/wav2vec2-base-arabic-finetuned-continued-v2 on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.5914
- Wer: 49.8920
- Cer: 14.9775
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: constant_with_warmup
- lr_scheduler_warmup_steps: 50
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.3845 | 1.0 | 884 | 0.6069 | 51.7863 | 15.5752 |
| 0.3632 | 2.0 | 1768 | 0.6077 | 50.4736 | 15.1016 |
| 0.3547 | 3.0 | 2652 | 0.5914 | 49.8920 | 14.9775 |
Framework versions
- Transformers 4.52.4
- Pytorch 2.6.0+cu124
- Datasets 2.18.0
- Tokenizers 0.21.2