arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.12247cs.AI
收藏

基于大型语言模型的轻量级特定领域问答模型的增强微调

Enhanced Fine-Tuning of Lightweight Domain-Specific Q&A Model Based on Large Language Models

Nankai University(南开大学);Microsoft(微软);Tsinghua University(清华大学);China Mobile Research Institute(中国移动研究院)

Shenglin Zhang, Pengtian Zhu, Minghua Ma, Jiagang Wang, Yongqian Sun, Dongwen Li, Jingyu Wang, Qianying Guo, Xiaolei Hua, Lin Zhu, Dan Pei

AI总结:

本文提出Self-Evolution框架,通过多轮迭代微调轻量级开源LLMs并筛选高价值知识,在保护隐私和节约资源的同时,显著提升特定领域问答性能并成功应用于实际运维。

摘要:

大型语言模型(LLMs)擅长通用问答(Q&A),但由于缺乏特定领域知识,在专业领域往往表现不佳。商业公司在使用LLMs进行微调时面临隐私保护和资源限制的双重挑战。本文提出了一种新颖的框架Self-Evolution,旨在通过利用轻量级开源LLMs进行多轮迭代微调来解决这些问题。为提升迭代微调的效率,Self-Evolution采用了一种策略,在迭代过程中筛选并强化具有较高价值的知识。我们在Qwen1.5-7B-Chat上使用了来自中国移动的4,000份包含丰富领域知识的文档进行Self-Evolution实验,在特定领域问答评估中,其性能得分比Qwen1.5-7B-Chat高出174%,甚至比Qwen1.5-72B-Chat高出22%。Self-Evolution已在中国移动的日常运维中部署了117天,提高了定位告警、修复问题和查找相关报告的效率,平均效率提升超过18.6%。此外,我们在https://github.com/Zero-Pointer/Self-Evolution发布了Self-Evolution框架代码。

英文摘要:

Large language models (LLMs) excel at general question-answering (Q&A) but often fall short in specialized domains due to a lack of domain-specific knowledge. Commercial companies face the dual challenges of privacy protection and resource constraints when involving LLMs for fine-tuning. This paper propose a novel framework, Self-Evolution, designed to address these issues by leveraging lightweight open-source LLMs through multiple iterative fine-tuning rounds. To enhance the efficiency of iterative fine-tuning, Self-Evolution employ a strategy that filters and reinforces the knowledge with higher value during the iterative process. We employed Self-Evolution on Qwen1.5-7B-Chat using 4,000 documents containing rich domain knowledge from China Mobile, achieving a performance score 174% higher on domain-specific question-answering evaluations than Qwen1.5-7B-Chat and even 22% higher than Qwen1.5-72B-Chat. Self-Evolution has been deployed in China Mobile's daily operation and maintenance for 117 days, and it improves the efficiency of locating alarms, fixing problems, and finding related reports, with an average efficiency improvement of over 18.6%. In addition, we release Self-Evolution framework code in https://github.com/Zero-Pointer/Self-Evolution.

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。

↑