Mirror of the documentation in the DetFill-with-DHT repository (source file CONTRIBUTING.md); links to files and release downloads point at GitHub, and the hintauc library is available from PyPI.

Contributing

Thank you for your interest in this project. Contributions of every size are welcome: bug reports, documentation fixes, new metrics or segmenters for the hintauc library, and reproduction reports.

Reporting issues

  • Use the GitHub issue tracker. Say what you ran (command line or code), what you expected, what happened, and your environment (OS, Python, PyTorch/CUDA versions, GPU or CPU).
  • For a reproduction that does not match the paper or the authors' reference numbers, attach the metric JSON or the console output of the script you used (replicability/run.sh, reproduce/examples/run_examples.sh, ...).

Pull requests

  1. Fork the repository and create a branch from main.
  2. Make your change. Keep the library (hintauc/) free of hard-coded paths and of dependencies that are not available under a free licence.
  3. Add or update tests in tests/ for behaviour changes in hintauc/, and run them: bash pip install -e ".[perceptual]" pytest pytest tests
  4. If the change affects the deterministic hint generator or the metrics, state in the pull request whether the released numbers change (python reproduce/scripts/A1_tables_from_released_metrics.py and the example suite are the reference checks).
  5. Update the documentation that describes what you changed (README.md, docs/, or the README of the component).
  6. Open the pull request with a short description of the motivation and of how you tested it.

Style

  • Python 3.9 compatible, PEP 8, snake_case; docstrings for public functions of hintauc.
  • No files that reveal private data, credentials or machine-specific paths.

Licence

By contributing you agree that your contributions are licensed under the MIT License of this repository (see LICENSE). Third-party code must keep its own licence notice and be listed in THIRD_PARTY_NOTICES.md.