RAG / 检索增强生成

PassiveQA: A Three-Action Framework for Epistemically Calibrated Question Answering via Supervised Finetuning

PassiveQA: 一个基于监督微调的三动作框架，用于通过监督微调进行知识校准的问题回答

Madhav S Baidya

机构 * Indian Institute of Technology (BHU) Varanasi（印度理工学院（巴纳拉斯印度教大学）瓦拉纳西分校） 纠错

专题命中 知识库问答 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL、cs.AI

AI总结 本文提出PassiveQA框架，通过监督微调解决信息不全时的问题回答，提升宏F1和退避召回率，减少幻觉。

Comments 32 pages, 4 figures. Includes experiments on four QA datasets and a knowledge graph-based finetuning pipeline. Code available at: https://github.com/MadsDoodle/PassiveQA

Talk to Right Specialists: Iterative Routing in Multi-agent Systems for Question Answering

与正确专家对话：多智能体系统中用于问答的迭代路由

Feijie Wu, Zitao Li, Fei Wei, Yaliang Li, Bolin Ding, Jing Gao

机构 * Purdue University（普渡大学） ； Alibaba（阿里巴巴） ； Zoom Communication（Zoom通信） 纠错

专题命中 知识库问答 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL、cs.AI

AI总结 本文提出RIRS框架，通过迭代路由解决多智能体问答中用户无法确定咨询哪个专家及复杂问题需多专家证据的问题，提升回答准确性和效率。

Comments Differences between v1 & v2: The algorithm name of the first version is RopMura, which decomposes a multi-hop query into several simple subqueries, and a question selector selects one of the subqueries to answer. In the second version, the name is updated to RIRS, which directly routes a query to the appropriate agents, regardless of whether the query is single-hop or multi-hop

Beyond Code Snippets: Benchmarking LLMs on Repository-Level Question Answering

超越代码片段：在仓库层面评估大语言模型的问答任务

Yoseph Berhanu Alebachew, Hunter Leary, Swanand Vaishampayan, Chris Brown

机构 * Department of Computer Science, Virginia Tech（弗吉尼亚理工大学计算机科学系） 纠错

专题命中 知识库问答 ：retrieval-augmented generation(abstract)；分类 cs.AI

AI总结 本文提出StackRepoQA数据集，用于评估大语言模型在多项目仓库层面的问答能力，通过对比不同配置下的性能，揭示了模型在处理仓库级程序理解时的局限性。