📄 world_models

🧠 NeurIPS2026 · 1 paper notes

StarWM: Self-Supervised Trained Attention Routing for Robust World Models

StarWM uses self-supervised dynamics signals to determine spatial attention, then constrains reconstruction through stop-gradient barriers and a dual-stream decoder so that world models retain predictable entities under randomized video distractions; coherent video backgrounds still confuse default routing, while the reward-guided StarWM-R variant improves cross-condition performance.