ServiceNow-AI/R1-Distill-SFT

这是一个从R1-32b模型中提取的数百万规模的数据集，用于支持开源社区。数据集分阶段开源，包括生成的样本、未过滤/未验证的数据集以及后续的过滤和验证版本。数据集包含ID、重注释的助手内容、问题、来源、解决方案等字段，并针对每个提示采样一个响应。

This is a millions-scale dataset distilled from the R1-32b model, supporting the open-source community. The dataset is open-sourced in stages, including generated samples, unfiltered/unverified dataset, and subsequent filtered and verified versions. The dataset includes fields such as ID, reannotated assistant content, problem, source, solution, and samples one response per prompt.