Fetching the paper…

DreamingV2: Reinforcement Learning with Discrete World Models without Reconstruction · Around