Key points are not available for this paper at this time.
检索增强生成(RAG)为大型语言模型(LLMs)引入了一种新范式,帮助解决知识密集型任务。然而,当前的RAG模型将LLMs定位为被动的知识接受者,从而限制了它们学习和理解外部知识的能力。本文提出了ActiveRAG,一种创新的RAG框架,将被动知识获取转变为主动学习机制。该方法利用知识构建机制,通过将外部知识与先前获得或记忆的知识关联,来深化对外部知识的理解。随后,设计了认知枢纽机制,融合了思维链和知识构建的结果,从而校准LLMs的内在认知。实验结果表明,ActiveRAG优于以往的RAG模型,在问答数据集上提升了5%的表现。所有数据和代码均可在https://github.com/OpenMatch/ActiveRAG获取。
Building similarity graph...
Analyzing shared references across papers
Loading...
Zhipeng Xu
Zhenghao Liu
Yibin Liu
Building similarity graph...
Analyzing shared references across papers
Loading...
Xu等人(星期三)研究了该问题。
www.synapsesocial.com/papers/68e7845cb6db6435876f723c — DOI: https://doi.org/10.48550/arxiv.2402.13547
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: