Self-supervised Hierarchical Visual Reasoning with World Model

Yuanfei Xu, Lin Liu, Wengang Zhou, Mingxiao Feng, Houqiang Li · 2026

Read Self-supervised Hierarchical Visual Reasoning with World Model by Yuanfei Xu, Lin Liu, Wengang Zhou, Mingxiao Feng, Houqiang Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-17537.txt.

Description

Xu and colleagues present ResDreamer, a hierarchical world model for visual reasoning in 3D open-world settings with hostile rivals. Each higher layer rebuilds the residuals of the layer below, abstracting world motion step by step and enriching latent codes. Higher-level residuals steer lower-level predictions, so the model scales with only slowly growing cross-layer talk. Trained purely self-supervised, it achieves leading sample and parameter efficiency in experiments.

Catalog record

CREATORYuanfei Xu, Lin Liu, Wengang Zhou, Mingxiao Feng, Houqiang Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43688
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2605.17537

Browse the full library · About the Tower of Babel