Image-to-Text
MLX
mlx-vision
ocr
apple-silicon
speculative-decoding
dspark
deepseek-ocr
glm-ocr
vision-language-model
Instructions to use will702/mlx-vision with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use will702/mlx-vision with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir mlx-vision will702/mlx-vision
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
| cff-version: 1.2.0 | |
| title: "mlx-vision" | |
| message: "If you use mlx-vision, please cite it as below." | |
| type: software | |
| authors: | |
| - family-names: "Willson" | |
| given-names: "Gregorius" | |
| alias: "will702" | |
| repository-code: "https://github.com/will702/mlx-vision" | |
| url: "https://huggingface.co/will702/mlx-vision" | |
| abstract: >- | |
| DSpark-style speculative decoding for OCR vision-language models on | |
| Apple Silicon via MLX. Speeds up DeepSeek-OCR-2, GLM-OCR, and Unlimited-OCR | |
| with lossless target verification when neural drafters are available. | |
| keywords: | |
| - mlx | |
| - ocr | |
| - speculative-decoding | |
| - dspark | |
| - apple-silicon | |
| - deepseek-ocr | |
| license: MIT | |
| version: 0.1.0 | |
| date-released: "2026-07-28" | |