CS858 Wiki

Per-paper reading companions for CS858 (Trustworthy Machine Learning). Each primary-reading paper gets one page that lowers the prerequisite load for reading it and surfaces the open questions it sits inside. Pages deliberately do not summarize the papers' own findings. That is what reading the paper is for.

Students — start here

Open the wiki-f26/ folder and begin at wiki-f26/README.md. Every page renders here with working links; follow them to the paper pages and the shared concept pages. You do not need any special tool; it is plain Markdown.

What's in this repo

• wiki-f26/ — the content: paper pages and shared concept pages, linked with plain relative Markdown links.
• docs/ops/ — the operational playbooks (how the pages are produced).
• .claude/commands/ — the /generate-paper-summary slash command.
• scripts/ — an arXiv fetcher, a PDF reader, and a link checker.

AGENTS.md is the operator contract for whoever maintains the wiki.

Maintainer setup

uv sync                                          # create the env, install tooling
uv run python3 scripts/fetch-arxiv.py <url|id>   # download a paper from arXiv
uv run python3 scripts/read-pdf.py <pdf>         # read a paper
uv run python3 scripts/check-links.py            # validate all links resolve

Adding a paper

/generate-paper-summary <arXiv-URL | arXiv-ID | pdf-path>

See docs/ops/generate-paper-summary.md for the full procedure and page schema.