arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.02333cs.CLcs.AI
收藏

关键点驱动的数据合成及其对数学推理的增强

Key-Point-Driven Data Synthesis with its Enhancement on Mathematical Reasoning

Microsoft(微软公司)

Yiming Huang, Xiao Liu, Yeyun Gong, Zhibin Gou, Yelong Shen, Nan Duan, Weizhu Chen

AI总结:

提出KPDDS关键点驱动数据合成框架,构建含80余万问答对的KPMath及KPMath-Plus数据集,使Qwen1.5-72B在GSM8K和MATH上取得领先成绩。

摘要:

大型语言模型(LLM)在复杂推理任务中展现出巨大潜力,但其性能往往因缺乏高质量、以推理为核心的训练数据集而受限。为应对这一挑战,我们提出了 Key-Point-Driven Data Synthesis(KPDDS,关键点驱动数据合成),这是一种新型数据合成框架,通过利用真实数据源中的关键点和示例实践来合成问答对。KPDDS 通过严格的质量控制和强大的可扩展性,确保生成新颖问题。因此,我们推出了 KPMath,这是一个专为数学推理定制的大规模合成数据集,包含超过80万个问答对。利用 KPMath 并以额外的推理密集型语料库进行扩充,我们构建了全面的 KPMath-Plus 数据集。在 KPMath-Plus 上微调的 Qwen1.5-72B 模型在 GSM8K 上达到87.0%的 PASS@1 准确率,在 MATH 上达到58.3%,在多个数学推理数据集上超越了7B至70B规模的竞争者以及 GPT-4 等最佳商业模型。

英文摘要:

Large language models (LLMs) have shown great potential in complex reasoning tasks, yet their performance is often hampered by the scarcity of high-quality and reasoning-focused training datasets. Addressing this challenge, we propose Key-Point-Driven Data Synthesis (KPDDS), a novel data synthesis framework that synthesizes question-answer pairs by leveraging key points and exemplar practices from authentic data sources. KPDDS ensures the generation of novel questions with rigorous quality control and substantial scalability. As a result, we present KPMath, an extensive synthetic dataset tailored for mathematical reasoning, comprising over 800K question-answer pairs. Utilizing KPMath and augmenting it with additional reasoning-intensive corpora, we create the comprehensive KPMath-Plus dataset. The Qwen1.5-72B model, fine-tuned on KPMath-Plus, achieves 87.0% PASS@1 accuracy on GSM8K and 58.3% on MATH, surpassing competitors in the 7B to 70B range and best commercial models like GPT-4 across multiple math reasoning datasets.

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。

↑