Reproduce “Zero-Shot Coordination among LLM 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
Adrian Hayler, Shashank Reddy Chirra, Andrei Lupu, Johannes Forkel, Bidipta Sarkar, Siheng Feng, Jakob Foerster
The paper tests whether a shared natural-language rule helps independently built LLM agents cooperate without agreeing on signal conventions. The hand-crafted rule warns against relying on arbitrary conventions and is added to agent prompts. In Cat & Dog, all 16 pairings averaged 5.00 points, the guaranteed fallback return; this result is limited to that game and evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.