When Should Multi-Round RAG Stop? Structured Stopping Judgments and Retrieval Reduction in Search-R1

  • 类型:arxiv
  • 标识:2608.13237
  • 链接:http://arxiv.org/abs/2608.13237v1
  • 主分类:rag
  • 形态:application
  • 被引:0
  • 被引来源:Semantic Scholar
  • S2被引:0
  • 影响力被引:0
  • TLDR:This work adapts S2G-RAG's structured sufficiency-and-gap judgment to a frozen Search-R1 pipeline and trains a Qwen3.5-2B judge on 3,009 states from 900 disjoint HotpotQA questions to reduce retrieval while broadly preserving answer accuracy.
  • 待LLM分类:否
  • 标题中文:多轮 RAG 何时停止?Search-R1 中的结构化停止判断与检索缩减
  • TLDR中文:本文将 S2G-RAG 的结构化充分性-缺口判断适配到冻结的 Search-R1 流程中,并在来自 900 个不相交 HotpotQA 问题的 3009 个状态上训练了一个 Qwen3.5-2B judge,以减少检索次数同时广泛保持答案准确性。
  • 来源文件
  • /inbox/tom/_candidates/2026-08-14-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-08-15-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-08-16-agent-rag-longcontext-candidates.json
  • [S2 enrich]