作者位次 #4/8ACL 2026Findings· Findings of ACL
B ayesian Active Learning with G aussian Processes Guided by LLM Relevance Scoring for Dense Passage Retrieval
Junyoung Kim · Anton Korikov · Jiazhou Liang · Justin Cui · Yifan Simon Liu · Qianfeng Wen · Mark Zhao · Scott Sanner
摘要
While Large Language Models (LLMs) exhibit exceptional zero-shot relevance modeling, their high computational cost necessitates framing passage retrieval as a budget-constrained global optimization problem. Existing approaches passively rely on first-stage dense retrievers, which leads to two lim...