Skip to content
English

Switch theme

arXiv 2610.26811 · cs.CV · 2026-10-19

State World Models: Execute the Rules, Generate the Pixels

Chan Yan Ki, Yanran Zhang, Wenzhao Zheng, Xiaofeng Wang, Zheng Zhu, Lei Chen, Jie Zhou, Jiwen Lu

Tsinghua-IVG

In latest · #308 Added on 2026-10-25

One-line takeaway

Tsinghua's Stateful Harness separates world evolution from rendering: it maintains a rule-governed world state, renders events with a frozen video model, and updates state from each clip, so world models obey persistent, invisible rules.

清华的 Stateful Harness 把世界演化和视频渲染分开:按规则维护世界状态,用冻结的视频模型渲染每个事件,再以生成片段更新状态,让交互式世界模型遵守看不见的持久规则。

17HF upvotes
22GitHub stars
0Citations (Semantic Scholar)
42.4Buzz score

Read on arXiv PDF Hugging Face discussion Code Project page

About this data

Metadata from the arXiv API (verified); upvotes from Hugging Face Daily Papers; citations from Semantic Scholar. Last refreshed 2026-10-26.

The takeaway is written by AIMap from the abstract; the abstract itself is not reproduced — read the original.

Data updated: Latest AI papers(UTC+8)