超越置信度：基于检索 grounding 的多轮搜索智能体测试时缩放

中文摘要 AI 辅助

基于置信度的投票通过 token 对数概率等内部信号对并行 LLM 推理路径进行加权聚合，已在单轮推理中得到广泛研究。然而，现代大语言模型（LLM）日益成为检索并依赖外部文档的多轮搜索智能体。本文表明，基于置信度的投票在多轮场景下迁移效果不佳，其根本原因是复制膨胀：当检索到的文档被添加到智能体的上下文时，从这些文档复制的 token 的对数概率会系统性地膨胀，这会拉平每个问题内的置信度分数，削弱加权投票的效果。为解决该问题，我们提出检索接地投票（Retrieval-Grounded Voting，RGV），通过每个推理路径的最终答案与其检索文档之间的词汇重叠度对其进行评分。由于在受污染的上下文之外计算信号，RGV 避开了 token 对数概率和额外的 LLM 调用。在四个搜索智能体基准和五个 LLM 上，RGV 的表现始终优于基于置信度的投票，准确率提升最高达 5.4%，在少数正确问题（8 条推理路径中仅 1-2 条包含正确答案）上的表现提升达 35%。

英文摘要

Confidence-based voting aggregates parallel LLM rollouts by weighting each with internal signals such as token log probabilities, and has been actively studied for single-turn reasoning. However, modern LLMs increasingly act as multi-turn search agents that retrieve and condition on external documents. In this paper, we show that confidence-based voting transfers poorly to this multi-turn setting, and identify the underlying failure reason as copy inflation: when retrieved documents are appended to an agent's context, tokens copied from those documents receive systematically inflated log probabilities. This flattens confidence scores within each question and weakens the resulting weighted vote. To address this issue, we propose Retrieval-Grounded Voting (RGV), which scores each rollout by the lexical overlap between its final answer and the documents it retrieved. By computing the signal outside the contaminated context, RGV sidesteps both token log probabilities and additional LLM calls. Across four search-agent benchmarks and five LLMs, RGV consistently outperforms confidence-based voting, with gains of up to +5.4% accuracy and +35% on minority-correct questions, where the correct answer appears in only 1-2 of 8 rollouts.

发表机构

• University of Southern California（南加州大学）
• Pohang University of Science and Technology (POSTECH)（浦项科技大学）
• Tsinghua University（清华大学）
• Carnegie Mellon University（卡内基梅隆大学）

机构由 AI 辅助整理，请以论文原文为准。

补充信息

备注

Accepted to EMNLP 2026 Findings