Group-Evolving Agents: Open-Ended Self-Improvement via Experience Sharing

This paper introduces Group-Evolving Agents (GEA), a new paradigm for open-ended self-improvement where a group of agents acts as the fundamental evolutionary unit, enabling explicit experience sharing and reuse. GEA significantly outperforms state-of-the-art self-evolving methods on challenging coding benchmarks and demonstrates consistent transferability and robustness.