Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
46 lines (33 loc) · 1.56 KB

CONTRIBUTING.md

File metadata and controls

46 lines (33 loc) · 1.56 KB
Raw

Contributing to TurboQuant

We welcome contributions! Here's how to get started.

Development Setup

git clone https://github.com/Firmamento-Technologies/TurboQuant.git
cd TurboQuant
pip install -e ".[dev]"
pytest tests/ -v

What We're Looking For

  • Performance improvements — Faster quantization/search, SIMD, batch operations
  • New bit-packing — Pack 4-bit codes into uint8 (2 values per byte)
  • GPU acceleration — Optional CuPy/PyTorch backend
  • FAISS adapter — Use TurboQuant as a FAISS custom index
  • Benchmarks — More datasets, higher dimensions, comparison with ScaNN/Annoy
  • Documentation — Tutorials, API docs, mathematical explanations

Guidelines

  1. Tests required — All PRs must include tests. Run pytest tests/ -v before submitting.
  2. Pure Python — Core algorithms must work with NumPy only. GPU/C++ extensions go in optional modules.
  3. Type hints — All public functions must have type annotations.
  4. Docstrings — Google-style docstrings for all public classes and methods.

Code Style

  • Python 3.10+ features (type unions with |, etc.)
  • NumPy-style array operations (avoid Python loops on vectors)
  • Keep modules focused — one responsibility per file

Submitting

  1. Fork the repo
  2. Create a feature branch (git checkout -b feat/my-feature)
  3. Write tests first, then implementation
  4. Run pytest tests/ -v and ensure all pass
  5. Submit a PR with a clear description

License

By contributing, you agree that your contributions will be licensed under the Apache License 2.0.