OSDC11 was launched today in Canberra. With a small team of volunteers, the conference has managed to pull together some 250 participants, sponsors, and talented speakers for the 8th year running.
Datasets and models now have their own pages, cards, file lists and a data viewer, much like Hugging Face — but every file is a link to where its publisher hosts it. glitchdata catalogues, previews and counts; it never keeps a copy.
Paste a Hugging Face ID or address on the New dataset or New model page and glitchdata builds a draft from it — card, licence, tags, key facts and every file as a link back to Hugging Face. Nothing is downloaded.
A new Adopt button adds any live dataset to My Datasets without taking ownership, and the redesigned dashboard brings your datasets, models and courses together — with a Continue button that picks up at your next lesson.
The course catalogue has been rebuilt around AI: thirteen free courses and 52 lessons, from how machine learning works to building and evaluating LLM agents — several of them hands-on with datasets and models from the hub.