Universal Models archive

A static restoration of Stanford's 2021 Universal Models website, originally called Foundations of Responsible Universal Models (FORUM). It preserves the site's original wording, design, and workshop announcement. Every page has a notice identifying it as an archived copy, not the current Stanford site.

The starting point was the July 28, 2021 workshop snapshot.

What was recovered

• docs/workshop.html: the July 28, 2021 workshop page.
• docs/index.html: the August 3, 2021 homepage.
• Styles, scripts, logos, and workshop artwork: August 24, 2021 captures of the original domain.
• Missing shared styles, icon fonts, and the Stanford panorama: August/September 2021 captures of crfm.stanford.edu. Several shared files have identical archive hashes on both domains. No later CRFM page text was substituted.

No original People page capture was found. people.html explains that gap instead of inventing a roster. Missing mobile navigation logos use the recovered FORUM logo.

Bootstrap 5.0.1 and jQuery 3.6.0 were downloaded from the original pinned CDN URLs and verified against the integrity hashes in the HTML. Google Fonts files are current downloads of the original Manrope and DM Serif Display families, not verified 2021 binaries. Display assets are served locally. Original external links remain external. Unused video/gallery styles still reference several unarchived images.

Preview

From this directory:

python -m http.server 8765 --bind 127.0.0.1 --directory docs

Open http://127.0.0.1:8765/ or http://127.0.0.1:8765/workshop.html. No build step is needed.

Publishing

In the repository's GitHub Pages settings, select Deploy from a branch, branch main, and folder /docs.

docs/CNAME specifies universalmodels.fast.ai. Its DNS-only Cloudflare CNAME points to fastai.github.io. Configure the same custom domain in the repository's Pages settings.

Recovery helpers and checks

recover.py contains the small helpers used interactively with the Python wayback package. It is not a complete recovery script. test_site.py checks local page links, assets, font/style dependencies, and archive notices.

With pytest and beautifulsoup4 installed, run:

pytest -q

Local recovery logs are stored in meta/ and are excluded from git. Original Stanford content and third-party assets retain their respective ownership and licenses.