RyanYr/DAPO-Math-17k-pg-dapo-qwen3-4B-Base-mbs128-n4_shuffled

该数据集是一个包含问题-响应对的结构化数据集，用于自然语言处理任务。它包含以下特征：problem（问题文本）、response（响应文本）、uid（唯一标识符）、answer（答案文本）、data_source（数据来源）和comment（评论，其中包含step步骤信息）。数据集仅提供训练拆分，包含122,880个示例，总大小约为841MB，适用于模型训练和评估。This dataset is a structured collection of problem-response pairs designed for natural language processing tasks. It includes features such as problem (problem text), response (response text), uid (unique identifier), answer (answer text), data_source (data source), and comment (comment with a step field). The dataset only provides a training split with 122,880 examples and a total size of approximately 841MB, suitable for model training and evaluation.