sungyub/eurus-2-math-verl

这个数据集包含283,612个数学推理问题，经过处理转换为VERL格式，用于奖励模型和强化学习训练。数据集专注于高质量的数学问题验证和奖励模型，适用于数学问题解决的强化学习应用。

This dataset contains 283,612 mathematical reasoning problems, processed and converted to the VERL format for reward modeling and reinforcement learning training. The dataset focuses on high-quality verification and reward modeling for mathematical problem-solving in reinforcement learning applications.