Reproduce “A Unified Arena for LLM Agents across Diverse Eurogame” 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

Fengrui Zhang, Xiang Li, Ruomeng Ding, Jiayi Yuan, Yuchen Huang, Zhaoyang Han, Xiaoliang Wang, Daoyuan Chen and 1 more

DECE evaluates complete LLM agent configurations across ten games. Its shared interfaces standardize observations and legal actions, while rule-enforcing engines run seat-balanced tournaments and record outcomes and resource use. Rankings formed on nine games show mean Kendall tau b of 0.664 and 83.9% pair-order accuracy against the held-out game. However, rankings and game-specific profiles describe only tested configurations and games.

Summary by alphaXiv.

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