优化推理型检索代理的计算分配，发现重排阶段投入更多算力更有效。
Compute Allocation for Reasoning-Intensive Retrieval Agents
- 重点提升重排阶段模型能力与候选池深度，查询扩展收益递减。
- 更强模型使NDCG@10提升7.5分，候选池从10扩至100使效果提高21%。
- 推理思考环节增益微弱，适合资源有限时优先保障重排阶段。
随着智能体运行时间延长，其记忆存储持续增长，检索成为获取相关信息的关键。许多查询需依赖推理型检索，即查询与相关文档间的关联隐含，需推理推断才能连接。现有基于大模型的管道通过查询扩展和候选重排来应对，但带来显著推理开销。本文基于BRIGHT基准和Gemini 2.5模型族，研究推理型检索管道中的计算分配问题，调整查询扩展与重排阶段的模型容量、推理思考深度及候选池规模。结果表明，重排阶段受益于更强模型（NDCG@10提升+7.5）和更深候选池（$k$从10增至100时提升21%），而查询扩展在轻量模型后收益递减（强模型仅比弱模型提升+1.1 NDCG@10）。推理思考在任一阶段均未带来明显改进。因此建议将算力集中投向重排阶段，而非均匀分配于各环节。
原文摘要 · Abstract (English)
As agents operate over long horizons, their memory stores grow continuously, making retrieval critical to accessing relevant information. Many agent queries require reasoning-intensive retrieval, where the connection between query and relevant documents is implicit and requires inference to bridge. LLM-augmented pipelines address this through query expansion and candidate re-ranking, but introduce significant inference costs. We study computation allocation in reasoning-intensive retrieval pipelines using the BRIGHT benchmark and Gemini 2.5 model family. We vary model capacity, inference-time thinking, and re-ranking depth across query expansion and re-ranking stages. We find that re-ranking benefits substantially from stronger models (+7.5 NDCG@10) and deeper candidate pools (+21% from $k$=10 to 100), while query expansion shows diminishing returns beyond lightweight models (+1.1 NDCG@10 from weak to strong). Inference-time thinking provides minimal improvement at either stage. These results suggest that compute should be concentrated on re-ranking rather than distributed uniformly across pipeline stages.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。