This is a research/educational project without formal release branches.
Only the latest commit on main is supported.
This project trains models on local image data and has no network-facing components, so the attack surface is limited (e.g. malformed inputs during data loading). If you find a security issue anyway:
- Please do not open a public GitHub issue for it.
- Use GitHub's private vulnerability reporting feature on this repository, or contact the maintainer directly.
- Include steps to reproduce and the potential impact.
We'll aim to acknowledge reports within a week.