arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.09260cs.CLcs.CE
收藏

JaFIn:日语金融指令数据集

JaFIn: Japanese Financial Instruction Dataset

Faculty of Information Science;Technology Hokkaido University;School of Engineering The University of Tokyo 0000-0001-8519-5617;Technology Hokkaido University 0000-0001-5030-625X;Technology Hokkaido University 0000-0003-1987-5336

Kota Tanabe, Masahiro Suzuki, Hiroki Sakaji, Itsuki Noda

AI总结:

本研究构建了日语金融指令数据集JaFIn,通过指令微调提升大语言模型在金融领域的领域适应性,实验证明金融专用模型优于原始模型。

摘要:

我们为日语金融领域的大语言模型(LLM)构建了一个指令数据集。随着语言模型(包括LLM)越来越普及,语言模型的领域适配受到更多关注。本研究通过指令微调证明了领域适配的有效性。为此,我们提出了一个名为JaFIn(日语金融指令数据集)的日语指令微调数据。JaFIn是基于多种数据源人工构建的,其中包括提供广泛金融知识的日本政府网站。然后,我们利用JaFIn对多个LLM进行指令微调,证明我们专门针对金融领域的模型比原始模型具有更好的领域适应性。所创建的金融专用LLM通过定量日语金融基准测试和定性响应比较进行了评估,结果显示其性能优于原始模型。

英文摘要:

We construct an instruction dataset for the large language model (LLM) in the Japanese finance domain. Domain adaptation of language models, including LLMs, is receiving more attention as language models become more popular. This study demonstrates the effectiveness of domain adaptation through instruction tuning. To achieve this, we propose an instruction tuning data in Japanese called JaFIn, the Japanese Financial Instruction Dataset. JaFIn is manually constructed based on multiple data sources, including Japanese government websites, which provide extensive financial knowledge. We then utilize JaFIn to apply instruction tuning for several LLMs, demonstrating that our models specialized in finance have better domain adaptability than the original models. The financial-specialized LLMs created were evaluated using a quantitative Japanese financial benchmark and qualitative response comparisons, showing improved performance over the originals.

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。

↑