Reproduce “MultiAgent-OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation” 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
Yuhan Sun, Jiaxin Zhao, Yongdong Luo, Yibo Hu, Meiguang Jin, Junfeng Ma, Weixun Wang, Weihang Pan and 1 more
MultiAgent-OPD trains a hierarchical question-answering team by first training a search-and-answer Teacher, then using its frozen token-level guidance for Searchers while the Leader and Answerer learn from outcomes. Across seven English open-domain QA datasets, it reaches 38.68% macro avg@4, versus 34.62% for Prompted Critic; this fixed-checkpoint comparison uses unmatched workflow budgets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.