Configuration Parsing
Warning:
In adapter_config.json: "peft.task_type" must be a string
Whisper Small AR - Mohammed Bakheet
This model is a fine-tuned version of KalamTech/arabic-whisper-large-v2-peft-fine-tuning on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1868
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: 0.001
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- 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: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.4715 | 0.8315 | 500 | 0.2189 |
| 0.1643 | 1.6619 | 1000 | 0.2025 |
| 0.1107 | 2.4923 | 1500 | 0.1884 |
| 0.0735 | 3.3226 | 2000 | 0.1896 |
| 0.0453 | 4.1530 | 2500 | 0.1890 |
| 0.0211 | 4.9845 | 3000 | 0.1868 |
Framework versions
- PEFT 0.15.2
- Transformers 4.52.3
- Pytorch 2.7.0+cu126
- Datasets 3.6.0
- Tokenizers 0.21.1
- Downloads last month
- 57
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Model tree for KalamTech/arabic-whisper-large-v2-peft-fine-tuning
Base model
openai/whisper-large-v2