2026/02/04 by Zhaotian Weng, Antonis Antoniades, Deepak Nathani +3 · 1 voice · 4 citations
Computer Science · #Artificial Intelligence in Games #Evolutionary Algorithms and Applications #Reinforcement Learning in Robotics #cs.AI
paper · pdf · doi:10.48550/arxiv.2602.04837
openalex publication_date 2026/02/04 · arxiv published 2026/02/04 · arxiv updated 2026/02/04 · openalex created_date 2026/02/07 · openalex updated_date 2026/07/28
Open-ended self-improving agents can autonomously modify their own structural designs to advance their capabilities and overcome the limits of pre-defined architectures, thus reducing reliance on human intervention. We introduce Group-Evolving Agents (GEA), a new paradigm for open-ended self-improvements, which treats a group of agents as the fundamental evolutionary unit, enabling explicit experience sharing and reuse within the group throughout evolution. Unlike existing open-ended self-evolving paradigms that adopt tree-structured evolution, GEA overcomes the limitation of inefficient utilization of exploratory diversity caused by isolated evolutionary branches. We evaluate GEA on challenging coding benchmarks, where it significantly outperforms state-of-the-art self-evolving methods (71.0% vs. 56.7% on SWE-bench Verified, 88.3% vs. 68.3% on Polyglot) and matches or exceeds top human-designed agent frameworks (71.8% and 52.0% on two benchmarks, respectively). Analysis reveals that GEA more effectively converts early-stage exploratory diversity into sustained, long-term progress, achieving stronger performance under the same number of evolved agents. Furthermore, GEA exhibits consistent transferability across different coding models and greater robustness, fixing framework-level bugs in 1.4 iterations on average, versus 5 for self-evolving methods.