库帕思高质量教育思维链（Chain-of-Thought）数据集-数学篇（上篇）

教育领域正式宣布开源高等教育阶段计算机、数学和物理的CoT数据集，立足服务教育智能化与大模型推理能力训练，聚焦学科核心知识与复杂问题解决，按照“教育场景适配”的要求，打造专业性、全面性、大规模、多题型、高精度的推理数据基座。

The education sector has officially announced the open-sourcing of a Chain-of-Thought (CoT) dataset covering computer science, mathematics and physics for higher education. Designed to support educational intelligence advancement and the training of large language model (LLM) reasoning capabilities, this dataset focuses on core disciplinary knowledge and complex problem-solving. Developed in line with the requirement of "education scenario adaptation", it is constructed as a professional, comprehensive, large-scale, diverse-question-type and high-precision reasoning-oriented data corpus.

搜集汇总

背景与挑战

背景概述

该数据集聚焦离散数学与高等数学核心知识，包含选择、填空、判断、解答等题型，旨在为AI教育工具提供逻辑推理与连续运算的思维样本，以提升模型解题能力。所有数据经过严格清洗与标注，采用JSON Lines格式，不仅提供标准答案，还包含由大语言模型生成的采样答案及其详细思考链，并通过自动化评估确保高质量。