arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.21030cs.IRcs.CL
收藏

学习何时思考:面向高效文本检索的效用引导自适应推理

Learning When to Think: Utility-Guided Adaptive Reasoning for Efficient Text Retrieval

Carnegie Mellon University(卡内基梅隆大学);University of Pittsburgh(匹兹堡大学)

Yingtao Luo, Kangyi Zhao, Shiyi Du, Yubo Li

AI总结:

该研究针对文本检索中推理过度导致的特异性不足问题,提出AURA框架,通过监督中间推理状态并训练效用估计器自适应调整推理深度,在BRIGHT和MS MARCO基准上实现了检索效率与性能的优化。

摘要:

基于推理的嵌入可通过逐步细化语义约束提升检索效果,但额外推理可能引入过于泛化或语义含噪的抽象,削弱检索的特异性。现有方法大多通过最终检索结果监督推理,对细化过程本身的指导有限。我们提出AURA框架,其根据中间推理状态区分相关文档与难区分干扰项的效果,直接监督中间推理状态;还在前缀条件嵌入上训练轻量效用估计器,用于推理时自适应调整推理深度。在推理密集型BRIGHT基准上,AURA-4B实现了34.7的平均nDCG@10,每查询使用2.8个推理步骤和96个生成令牌;在相同部署条件下,与全轨迹推理相比,自适应停止将延迟从4.6秒降至0.76秒。在MS MARCO上的结果进一步表明,这种专业化不会大幅降低常规域内检索的性能。

英文摘要:

Reasoning-conditioned embeddings can improve retrieval by progressively refining semantic constraints. However, additional reasoning may introduce over-generalized or semantically noisy abstractions that weaken retrieval specificity. Existing methods largely supervise reasoning through final retrieval outcomes, providing limited guidance over the refinement process itself. We propose AURA, a framework that directly supervises intermediate reasoning states according to how well they separate relevant documents from hard distractors. AURA further trains a lightweight utility estimator over prefix-conditioned embeddings to adapt reasoning depth at inference time. On the reasoning-intensive BRIGHT benchmark, AURA-4B achieves 34.7 average nDCG@10 while using 2.8 reasoning steps and 96 generated tokens per query. Under matched deployment conditions, adaptive stopping reduces latency from 4.6 to 0.76 seconds relative to full-trajectory reasoning. Results on MS MARCO further suggest that this specialization need not substantially degrade conventional in-domain retrieval.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑