Reproduce “MAC-Bench: Benchmarking Multi-Agent Collaboration via Interconnected Tasks” 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
Zerui Chen, Qinggang Zhang, Qichuan Liu, Zhishang Xiang, Zhimin Wei, Yunbo Tang, Zhuo Li, Jinsong Su and 1 more
MAC-Bench evaluates whether systems can complete interdependent work across software engineering and business operations. It builds tasks from reusable collaboration patterns and workflow structures, with dependencies and shared-resource constraints, while leaving agent organization open. Meta-Team achieves higher progress than ReAct with DeepSeek-Flash, 23.88% versus 21.88%, but resolves fewer tasks, 2 of 48 versus 3; no tested configuration exceeds 6.25% resolution.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.