构建合成指令数据集以提升领域特定大语言模型的推理能力：以日本金融领域为例
Constructing Synthetic Instruction Datasets for Improving Reasoning in Domain-Specific LLMs: A Case Study in the Japanese Financial Domain
AI总结：
本研究提出了一种通用方法，通过构建合成指令数据集提升领域特定大语言模型的推理能力，并在金融领域进行了验证。
摘要：
在将大语言模型适应特定领域时，实现领域专业知识和推理能力仍然是一个紧迫的挑战。本研究提出了一种通用方法，用于构建高质量的合成指令数据集，从领域特定的词汇开始。作为演示，我们应用了这种方法到金融领域，并构建了一个总计约95亿个标记的大型指令数据集，包含链式推理轨迹。评估结果证实了在金融基准测试中相对于基线模型的性能提升，证明了我们方法的有效性。我们还报告了推理轨迹长度对性能的影响及其限制。最后，我们在https://huggingface.co/nri-ai上开源了我们的模型和数据集。
英文摘要：
In adapting LLMs to specific domains, achieving both domain expertise and reasoning ability remains an urgent challenge. This study proposes a general method for constructing high-quality synthetic instruction data for any domain, starting from domain-specific vocabulary. As a demonstration, we applied this method to the financial domain and constructed a large-scale instruction dataset totaling approximately 9.5 billion tokens with Chain-of-Thought reasoning traces. Evaluation results confirmed performance improvements over baseline models on financial benchmarks, demonstrating the effectiveness of our approach. We also report findings on the impact of reasoning trace length on performance and its limitations. Lastly, we open-source our models and datasets on https://huggingface.co/nri-ai .
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。
