Reproduce “RSI-Router: Evolving Subtask-Level LLM Routing and Skills for Cost-Efficient Agents” 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
Hao Li, Hangfan Zhang, Zhiyao Cui, Chunjiang Mu, Yiqun Zhang, Bo Zhang, Danyang Jia, Shuyue Hu
RSI-router assigns small and large models to stages within long-horizon agent tasks and provides skills. During development, a proposal agent evolves stage definitions, model assignments, and skills, retaining routers; at deployment, a small model identifies the stage and selects its model. Across five benchmarks, performance exceeds the large-model baseline while inference cost falls 8.1–82.2%, but this measures cost, not latency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.