NuminaMath-1.5 Mathematical Reasoning Dataset

*This dataset supports online use.Click here to jump.

The NuminaMath-1.5 dataset was released by Project-Numina in 2025. This dataset is suitable for the fields of mathematics education and competition problems. It contains about 900k high-quality competition-level mathematics problems, and the solution to each problem is in the Chain of Thought (CoT) format. These problems are derived from Chinese high school mathematics exercises and American and International Mathematical Olympiad competition problems. This dataset adds problem metadata based on the v1 version, classifies the problem types, and introduces more manually verified data.