Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMs

  • 类型:arxiv
  • 标识:2607.25600
  • 链接:http://arxiv.org/abs/2607.25600v1
  • 主分类:rag
  • 形态:method
  • 被引:0
  • 被引来源:Semantic Scholar + OpenAlex
  • S2被引:0
  • OpenAlex被引:0
  • 影响力被引:0
  • TLDR:BeyondUncertainty first elicits a structured provisional answer and confidence estimate, then applies a model-specific threshold selected on held-out validation data and frozen before test evaluation, revealing a trade-off between more selective evidence acquisition and end-to-end token efficiency.
  • OpenAlex ID:W7171708250
  • OpenAlex DOI:10.48550/arxiv.2607.25600
  • DOI:10.48550/arxiv.2607.25600
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://doi.org/10.48550/arxiv.2607.25600
  • OpenAlex更新:2026-08-25
  • 待LLM分类:否
  • 标题中文:超越自我认知:在 LLM 推理与检索间传播不确定性
  • TLDR中文:BeyondUncertainty 首先引出结构化的临时答案与置信度估计,然后应用在 held-out 验证数据上选定并在测试评估前冻结的模型特定阈值,揭示了更具选择性的证据获取与端到端 token 效率之间的权衡。
  • 来源文件
  • /inbox/tom/_candidates/2026-07-29-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-07-30-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]