PACEvolve++ is introduced, an advisor-model reinforcement learning framework for test-time policy adaptation in evolutionary search agents, which outperforms the state-of-the-art evolutionary search framework with frontier models, achieving faster convergence and stabilizing test-time training during evolutionary searc...
Minghao Yan, Bo Peng, Benjamin Coleman et al.· arXiv.org· 4 citations
This work studies the problem of LLM-based descriptor generation, i.e., generating keyphrase-like natural language item representations with minimal constraints on downstream applications with little control over the generation space.
Zhou-Hang Xie, Bo Peng, Zhan-Kui He et al.· Conference on Empirical Meth...· 3 citations
A lightweight orchestration layer makes exploration explicit and programmable while leaving the underlying base agent unchanged, and a key insight is that accumulated discovery history can act as a replay simulator over the realized search space.
Tong Zheng, Xidong Wu, Zheng Zhang et al.· 13 citations
Progress-Aware Consistent Evolution (PACEvolve), a systematic framework for governing agent memory and search dynamics, maintains a global view of search momentum and achieves state-of-the-art results across complex evolutionary benchmarks.
Minghao Yan, Bo Peng, Benjamin Coleman et al.· arXiv.org· 22 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.