Qari-OCR
Arabic OCRBased on Qwen2-VL-2B. Best for Arabic text with diacritics.
Quick install (recommended)
- Install dependency:
- Run the downloader:
Files will be saved to C:\voidtranslate\models\qari-ocr.
If Hugging Face asks for access, create a read token and run huggingface-cli login.
Verify
C:\voidtranslate\models\qari-ocr\config.jsonUse in app
- Open the app.
- OCR method: Qari.
- Source language: Arabic (recommended).
Alternative install methods
Git LFS
git lfs install mkdir "C:\voidtranslate\models\qari-ocr" git clone https://huggingface.co/NAMAA-Space/Qari-OCR-v0.3-VL-2B-Instruct "C:\voidtranslate\models\qari-ocr"Browser download
- config.json, generation_config.json, preprocessor_config.json
- tokenizer.json, tokenizer_config.json, vocab.json, merges.txt
- model.safetensors.index.json + all model-*.safetensors
huggingface-cli
huggingface-cli login huggingface-cli download NAMAA-Space/Qari-OCR-v0.3-VL-2B-Instruct --local-dir "C:\voidtranslate\models\qari-ocr"Troubleshooting
- "Cannot find the requested files in the disk cache" — check that
config.jsonexists in the folder. - "Out of Memory" — needs ~4 GB VRAM; if not available, run on CPU.
- First run can be slow while the model warms up.









