Reproduce “MultiEcho: An Experimental Science of Learned Worlds” 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
Meng Zhu, Airui Zhang
MultiEcho measures how frozen world models respond to controlled changes in simulated physical conditions, comparing each model’s full output change with its own fitted prediction and the simulator’s physical effect. In a heat experiment, LeWM’s self-description error ratio was 0.556, while its physical-effect error ratio was 1.013. This local held-out result concerns the tested subject and intervention.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.