Yining Hong

Unsupervised environment latents adapt robotic world models to unseen physical properties and rules from just one interaction, drastically improving prediction and task success.

Multimodal models can achieve significant image generation improvements by learning from their own critiques during test-time compute, effectively acting as their own recursive self-improvement mechanism.

This work introduces MindTopo, a benchmark of topological intuition across five properties grounded in cognitive science and formal topology: continuity, separation, order, enclosure, and knots, which performs better on reasoning than on planning, and the best-performing model remains far below observed human performance.