Metamath-qwen2-math

该数据集包含了大约90万个数学问题，每个问题都附有解题思路（Chain-of-Thought，简称CoT）的解决方案，这些问题来源于多个渠道，包括MetaMathQA和NuminaMath-CoT。此外，该数据集还融入了多样化的非合成数据，并包含了一个难度级别模型，以便于问题的选择。规模上，数据集涵盖了90万道问题，任务重点在于长链式解题思路推理。

This dataset contains approximately 900,000 mathematical problems, each accompanied by a Chain-of-Thought (CoT for short) solution. These problems are sourced from multiple channels including MetaMathQA and NuminaMath-CoT. Additionally, the dataset incorporates diverse non-synthetic data and a difficulty-level model to facilitate targeted problem selection. In terms of scale, the dataset encompasses 900,000 problems, with its core task focusing on long-chain chain-of-thought reasoning.