MathFusionQA
收藏资源简介:
MathFusionQA数据集是由中国人民大学高灵人工智能学院、上海人工智能实验室等机构提出的一个数学问题数据集。该数据集通过三种融合策略(顺序融合、平行融合和条件融合)从原有数学训练集中构建问题对,生成新的数学问题。这些问题涵盖了数学知识的基本关系和组成方面,旨在通过增强数学问题之间的逻辑联系来提升大型语言模型的数学推理能力。MathFusionQA数据集共有60000个样本,用于训练和评估数学问题解决模型。
The MathFusionQA dataset is a mathematical problem dataset proposed by institutions including the Gaoling School of Artificial Intelligence at Renmin University of China and the Shanghai AI Laboratory. It constructs question pairs and generates new mathematical problems from the original mathematical training corpus via three fusion strategies: sequential fusion, parallel fusion, and conditional fusion. These problems cover the basic relational and compositional aspects of mathematical knowledge, aiming to improve the mathematical reasoning capabilities of large language models by strengthening the logical connections between different mathematical problems. The MathFusionQA dataset contains a total of 60,000 samples, which are used for training and evaluating mathematical problem-solving models.




