wav2vec2-base-arabic-finetuned-continued
This model is a fine-tuned version of Mohammadawad1/wav2vec2-base-arabic-finetuned on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.6754
- Wer: 0.5993
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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- 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: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8112 | 1.0 | 444 | 0.8405 | 0.6893 |
| 0.7405 | 2.0 | 888 | 0.7649 | 0.6513 |
| 0.6785 | 3.0 | 1332 | 0.7303 | 0.6312 |
| 0.6484 | 4.0 | 1776 | 0.7013 | 0.6163 |
| 0.5973 | 4.9899 | 2215 | 0.6754 | 0.5993 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 2.18.0
- Tokenizers 0.21.1