Reproduce “MOON: Learning When to Forget and How to Move On in Vision-and-Language Navigation” 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
Jingmeng Zhou, Seson Sun, Binyao Xu, Lingming Zhang, Hanshi Wang, Fudong Ge, Weiming Hu, Jin Gao and 1 more
MOON helps a vision-and-language agent recover after deviating from an indoor route. It detects trouble, clears accumulated visual history while retaining the instruction and current image, and hands control to a trained recovery adapter; the base policy remains frozen. Paired tests show clearing raises off-route success from 13.7% to 28.6% but lowers on-route success; MOON leads R2R-CE, not RxR-CE.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.