Reproduce “Learning to Coordinate with Anyone” 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

Lei Yuan, Lihe Li, Ziqian Zhang, Feng Chen, Tianyi Zhang, Cong Guan, Yang Yu, Zhi-Hua Zhou

Macop trains cooperative controllers by repeatedly evolving teammate groups that are capable, diverse, and incompatible with the current controller, then learning coordination with them. Its shared backbone and specialized output heads support adaptation and retention across partners. Across eight scenarios, Macop improved average performance over Finetune by 60.44%, but evaluation covered finite generated teammate groups.

Summary by alphaXiv.

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