[论文] Task-Adaptive Embedding Refinement via Test-time LLM Guidance

## 论文概要 **研究领域**: NLP **作者**: Ariel Gera, Shir Ashury-Tahan, Gal Bloch, Ohad Eytan, Assaf Toledo **发布时间**: 2026-05-12 **arXiv**: [2605.12487](https://arxiv.org/abs/2605.12487) ## 中文摘要 我们探索 LLM 引导的查询优化范式，将嵌入模型的可用性扩展到具有挑战性的零样本搜索和分类任务。我们的方法使用生成式 LLM 对少量文档的反馈来优化用户查询的嵌入表示，使嵌入能够实时适应目标任务。我们在最先进的文本嵌入模型上跨多种挑战性搜索和分类基准进行广泛实验。实证结果表明，LLM 引导的查询优化在所有模型和数据集上产生一致增益，在文献搜索、意图检测、关键点匹配和细微查询指令遵循方面相对提升高达 +25%。优化后的查询提高排名质量并在语料库上诱导更清晰的二元分离，使嵌入空间更好地反映每个临时用户查询的细微、任务特定约束。 ## 原文摘要 We explore the effectiveness of an LLM-guided query refinement paradigm for extending the usability of embedding models to challenging zero-shot search and classification tasks. Our approach refines the embedding representation of a user query using feedback from a generative LLM on a small set of documents, enabling embeddings to adapt in real time to the target task. We conduct extensive experiments with state-of-the-art text embedding models across a diverse set of challenging search and classification benchmarks. Empirical results indicate that LLM-guided query refinement yields consistent gains across all models and datasets, with relative improvements of up to +25% in literature search, intent detection, key-point matching, and nuanced query-instruction following. The refined queries... --- *自动采集于 2026-05-14* #论文 #arXiv #NLP #小凯