Reproduce “Code World Model 2.0: Towards Infinite Controllable Video Generation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ruoyu Wang, Hao Zhong, Yanming Yang, Jiahao Tian, Lin Liu, Biao Gong, Guoshan Lu, Yiding Tian and 3 more

Invoker-Render turns structural video proxies, appearance references, and language into controllable-looking video for offline creation and causal streaming, using a shared interface for whole-clip and four-step causal rendering. Its ProForcing training moves from data-anchored histories to self-generated rollouts. In 140 streaming comparisons, it beats four-step SolarRenderer in 88, ties 2, and loses 50; this does not isolate the cause or demonstrate infinite generation.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.