高质量学科数据集

COT-Academic Pro 是一款面向大模型后训练的高质量多模态多学科数据集，涵盖大学数学、高中数 学、高中竞赛题、物理、化学和生物等学科。通过多学科、多难度水平和高质量思维链数据，提升大模型 在 STEM(科学、技术、工程、数学)领域的解题准确率和推理能力。

COT-Academic Pro is a high-quality multi-modal and multi-disciplinary dataset designed for post-training of large language models. It covers disciplines such as college mathematics, high school mathematics, high school competition problems, physics, chemistry, and biology. Leveraging multi-disciplinary, multi-difficulty levels and high-quality chain-of-thought data, this dataset aims to enhance the problem-solving accuracy and reasoning capabilities of large language models in the STEM (Science, Technology, Engineering, Mathematics) domain.

搜集汇总

背景与挑战

背景概述

COT-Academic Pro是一个高质量学科数据集，专注于训练AI的严谨逻辑推理和多模态理解能力。它通过高级科学推理场景（如IMO级难题）和图像-文本整合场景（如电路图解析）来提升模型的思维灵活性与解释性。