Reproduce “HandWorldBench: A Unified Benchmark for Hand-Centric World Models” 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

Zhaolu Wang, Jiacheng Xu, Lihong Wang, Shuqing Wang, Jianjie Fang, Ziyou Wang, Xvyuan Liu, Rongze Tang and 7 more

HandWorldBench evaluates whether generated videos follow hand-motion cues and produce, preserve, and remember intended object states across human, simulated, and real dexterous hands. It uses 500 evaluation cases, measuring hand control separately from object-state outcomes. Hand2World scores 0.529 on Pose Control but 0.021 on State Progression and 0.000 on State Regulation; advantages vary by embodiment.

Summary by alphaXiv.

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