TuringQ
收藏资源简介:
TuringQ是由伊朗谢里夫理工大学和卡塔尔计算研究所联合创建的数据集,专门用于评估大型语言模型在计算理论中的推理能力。该数据集包含4,006个本科和研究生级别的问题-答案对,涵盖四个难度级别和七个核心理论领域。数据集的内容来源于全球29所顶尖大学的公开考试和作业解决方案,以及通过Claude 3 Sonnet模型生成的核心理论问题。创建过程包括精心挑选和生成问题,确保数据集的高质量和广泛覆盖。TuringQ的应用领域主要集中在提升大型语言模型在复杂计算推理任务中的表现,旨在解决计算理论中的深度理解和逻辑推理问题。
TuringQ is a dataset jointly created by Sharif University of Technology in Iran and the Qatar Computing Research Institute, specifically designed to evaluate the reasoning capabilities of large language models in computational theory. This dataset contains 4,006 question-answer pairs at undergraduate and graduate levels, covering four difficulty tiers and seven core theoretical domains. The content of the dataset is sourced from public exams and assignment solutions from 29 top-tier universities worldwide, as well as core theoretical questions generated using the Claude 3 Sonnet model. The dataset creation process involves meticulous selection and generation of questions to ensure its high quality and broad coverage. The primary application scenarios of TuringQ focus on enhancing the performance of large language models in complex computational reasoning tasks, aiming to address deep understanding and logical reasoning problems in computational theory.




