IPR-1: Interactive Physical Reasoner 事件

IPR-1: Interactive Physical Reasoner arXiv:2511.15407v4 Announce Type: replace-cross Abstract: Humans learn by observing, interacting with environments, and internalizing physics and causality. Here, we aim to ask whether an agent can similarly acquire human-like reasoning from interaction and keep improving with more experience. To study this, we introduce a Game-to-Unseen (G2U) benchmark of 1,000+ heterogeneous games that exhibit significant visual domain gaps. Existing approaches, including