ActiveUltraFeedback: Efficient Preference Data Generation using Active Learning 文章

ArXiv CS.CL2026-06-02NEWSen作者: Davit Melikidze, Marian Schneider, Jessica Lam, Martin Wertich, Ido Hakimi, Barna P\'asztor, Andreas Krause

查看原文 →

ActiveUltraFeedback: Efficient Preference Data Generation using Active Learning · 相关技术

相关技术

DELTAUCB DOUBLE REVERSE THOMPSON SAMPLING (DRTS)large language models reinforcement learning from human feedback