Reproduce “MoRSE: Task-Oriented Multi-Agent System with Mixture of Role-Subtask Experts” 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

Peiwen Li, Shiyang Zhang, Yangtian Zhang, Sizhuang He, David van Dijk, Rex Ying

MoRSE assigns agents reusable roles and instance-specific subtasks in a dependency-aware graph, then adapts a shared frozen model with role and subtask experts selected by a semantic router. On SRDD, its task structure shows lower output similarity than role-prompted MacNet. With trained experts, Qwen3-4B-Instruct reaches 86.25% Exec versus 72.50% for MoRSE_base on the test split. Evidence is limited to code-generation benchmarks.

Summary by alphaXiv.

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