Openness is graded, not claimed
Every model card scores four Model Openness Framework axes — weights, training data, training code, evaluation. Restrictive licenses are rejected in CI automatically.
Systems change through AI democratization
An open model & dataset registry that grades real openness, a free technology school, and a community where contributing is how you belong. No open-washing — every card, course, and governance rule in this platform lives in a public repo you can fork today.
Three commitments, enforced by the build system — not by a marketing page.
Every model card scores four Model Openness Framework axes — weights, training data, training code, evaluation. Restrictive licenses are rejected in CI automatically.
Courses built for four community personas, from first inference to publishing your own open model. Free, forever, no account required to learn.
Static site, client-side search, no tracking cookies, no user database. We cannot surveil you because there is no server to do it.
Multimodal generative AI for radiological analysis — detection, anatomical segmentation, and GAN-based image enhancement, trained with federated learning.
A hands-on 90-minute working session. Bring a model you want to document; leave with a merged pull request.
Every model card, course, and dataset here was added by someone through a pull request. Add yours — you don't need to know git.
Make your first contribution