arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2502.14271cs.CL
收藏

PaperHelper:基于知识的 LLM 问答论文阅读助手

PaperHelper: Knowledge-Based LLM QA Paper Reading Assistant

University of Minnesota, Twin Cities(明尼苏达大学双城分校)

Congrui Yin, Evan Wei, Zhongxing Zhang, Zaifu Zhan

AI总结:

本文提出 PaperHelper 论文阅读助手,结合 RAG、RAFT 与 RAG Fusion 降低 LLM 幻觉并支持文献批量管理和关系可视化,微调 GPT-4 API 后 F1 达 60.04、延迟 5.8 秒。

摘要:

在本文中,我们介绍了一款论文阅读助手 PaperHelper,这是一个功能强大的工具,旨在提升研究人员高效浏览和理解科学文献的能力。通过利用检索增强生成(RAG)框架,PaperHelper 有效减少了大型语言模型(LLM)中常见的幻觉,优化了准确、高质量知识的提取。RAFT 和 RAG Fusion 等先进技术的实现,显著提升了基于 LLM 的文献综述过程的性能、准确性和可靠性。此外,PaperHelper 具备用户友好的界面,支持批量下载文档,并使用 Mermaid 格式展示文档之间的结构关系。实验结果表明,基于微调 GPT-4 API 的 PaperHelper 取得了 60.04 的 F1 分数,延迟仅为 5.8 秒,在 F1 分数上比基础 RAG 模型高出 7%。

英文摘要:

In the paper, we introduce a paper reading assistant, PaperHelper, a potent tool designed to enhance the capabilities of researchers in efficiently browsing and understanding scientific literature. Utilizing the Retrieval-Augmented Generation (RAG) framework, PaperHelper effectively minimizes hallucinations commonly encountered in large language models (LLMs), optimizing the extraction of accurate, high-quality knowledge. The implementation of advanced technologies such as RAFT and RAG Fusion significantly boosts the performance, accuracy, and reliability of the LLMs-based literature review process. Additionally, PaperHelper features a user-friendly interface that facilitates the batch downloading of documents and uses the Mermaid format to illustrate structural relationships between documents. Experimental results demonstrate that PaperHelper, based on a fine-tuned GPT-4 API, achieves an F1 Score of 60.04, with a latency of only 5.8 seconds, outperforming the basic RAG model by 7\% in F1 Score.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑