Three things are next for PITF, in rough order:
A hub for models makes no sense without the data they train on. The datasets page is already up with preview data; a real index with search and cards comes next.
Publishing is currently anonymous and wide open — fine for a demo, wrong for a real community. Accounts will bring ownership: edit your model cards, delete your submissions, and keep a handle on your org namespace.
Placeholder stubs keep the demo fast, but a model hub that can't host models is a punchline. Real file uploads (with real sizes) are on the list.
That's the plan. Now back to the gym. 🥊