mmcontext#

Align embeddings across multiple modalities using context-aware embeddings at the sample level.

mmcontext is built upon the excellent sentence-transformers framework maintained by Hugging Face. By leveraging their comprehensive documentation and extensive capabilities for text embeddings, mmcontext enables you to efficiently generate multi-modal embeddings without reinventing the wheel.

Conceptual Diagram

Getting Started#

Please refer to our documentation, especially the API documentation. This project was built using the amazing scverse cookie cutter template—check it out!

Installation#

Ensure you have Python 3.10 or newer installed.

Currently, mmcontext is available for installation directly from the GitHub source:

pip install git+https://github.com/mengerj/mmcontext.git@main

# If sentence-transformers dependency is outdated, use:
pip install git+https://github.com/mengerj/sentence-transformers.git@master

(PyPI release coming soon!)

Contributing#

This package is under active development. Contributions and suggestions are very welcome—please open an issue to propose enhancements, report bugs, or discuss potential improvements.

Release Notes#

Review the latest changes in the Changelog.

Contact#

Encountered a bug or need help? Please use the issue tracker.

Citation#

If you find mmcontext useful for your research, please consider citing it as follows until an official publication is available:

@misc{mmcontext,
  author = {Jonatan Menger},
  title = {mmcontext: Multi-modal Contextual Embeddings},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub Repository},
  url = {https://github.com/mengerj/mmcontext}
}

mmcontext

Contents