AI Tinkerers Calgary Research Paper Club: LeWorldModel

Calgary''s paper club takes apart LeWorldModel (LeWM), described by its authors as the first joint-embedding predictive architecture that “trains stably end-to-end from raw pixels using only two loss terms”. The session is a deep dive on the architecture and its training regime rather than a summary read-through.

This one runs online, in the chapter''s Discord, 18:00–19:00 — despite being a Calgary chapter event, there is no room to turn up to. Space is limited to 150 and a call for presenters was open at the time of writing.