Seanie-lee/ThinkSafe-R1-Distill-1.5B-v3
收藏Hugging Face2025-10-31 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/Seanie-lee/ThinkSafe-R1-Distill-1.5B-v3
下载链接
链接失效反馈官方服务：
资源简介：
该数据集经过LlamaGuard过滤，包含了被拒绝的有害提示、生成拒绝的有害提示以及良性提示。在过滤之前，数据集共有39991个示例，过滤后共有38571个示例，移除了1420个有害示例，安全比为96.45%。
The dataset, after being filtered by LlamaGuard, includes harmful prompts that were refused, harmful prompts with generated refusals, and benign prompts. Initially, the dataset contained 39991 examples, and after filtering, it has 38571 examples, with 1420 harmful examples removed, resulting in a safe ratio of 96.45%.
提供机构：
Seanie-lee
