akahana/OpenThoughts-114k-math

这是一个经过过滤和元数据丰富的数据集，基于OpenThoughts-114k。它专注于数学内容，包含了原始的问题陈述、解决方案、消息对话、系统提示、生成的令牌数量以及正确性标签。数据集通过Math-Verify库验证了模型的答案正确性，并提供了每个问题的单一生成。该数据集适用于数学问题解答的模型训练和评估。

This is a filtered and metadata enriched dataset based on OpenThoughts-114k. It focuses on math content and includes the original problem statement, solution, message turns, system prompt, generated token count, and correctness label. The dataset has been verified for model answer correctness using the Math-Verify library, providing a single generation per question. It is suitable for training and evaluating models on math problem answering.