Important
This document is written both as a guide and example. Explanation parts are in Notes blockquotes and can simply be removed while building the repository.
Note
One short paragraph describing the purpose of the project.
Note
Whether your project is built around pip, conda, uv or another tool, provide here the necessary information for a full setup. Inform in the README about the basic setup needs (coding language and last version it is updated for, prefered OS, hardware and processing units).
To facilitate it for other users, preferably prepare an install that is usable on the various system softwares and common computer hardwares. If that isn't possible, try to document the installation required to make it work with different computers and systems, and if any isn't supported. You can also mention if the project has only been tested with a specific OS or processors, or for example if it works with different ones but is faster on a prefered one.
We recommend using uv as it is a fast and easy tool, enables building the code into its own package and gathers all dependencies into a single file for an easier install. Use the pyproject.toml file to declare the dependencies.
Here's how it can be presented:
This project uses uv for dependency management.
Here add more detail information about the install, as explained above
Install dependencies in a local venv with:
uv syncDependencies are listed in pyproject.toml.
Use uv add ... to add packages, and view uv docs for more info.
Run the main file with:
uv run python src/main.pyIf using the code as a package with internal relative dependencies, run
uv buildfrom project root to build the package and then run main with:
uv run python -m python_project_template.mainExplain structure:
.
├── LICENSE
├── pyproject.toml
├── README.md
└── src
└── python_project_template
└── main.py
Where data comes from, where it should live, and what is not committed.
How to run one minimal experiment.
Python version, uv.lock, seeds, hardware assumptions.
Explain how people can cite your work by filling in CITATION.cff with your info.
MIT license by default. Update your name and the year in LICENCE.
If you are a member of SFI Visual Intelligence and want to make changes, go ahead, it's open.