llm-semantic-router/mlcommons-ai-safety-synth
收藏Hugging Face2026-01-23 更新2026-02-07 收录
下载链接：
https://hf-mirror.com/datasets/llm-semantic-router/mlcommons-ai-safety-synth
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含12,000个合成的非安全提示，覆盖6个危害类别，旨在增强内容安全分类器的训练数据。每个类别包含2,000个平衡样本。危害类别包括非暴力犯罪、专业建议、隐私、无差别武器、自杀自残和选举错误信息。数据是通过few-shot prompting方法生成的，使用了NVIDIA AEGIS AI Content Safety Dataset 2.0的真实例子作为种子模式。数据集以JSONL格式存储，每个样本包含文本、类别和标签字段。
This dataset contains 12,000 synthesized unsafe prompts across 6 hazard categories, designed to augment training data for content safety classifiers. Each category contains 2,000 balanced samples. Hazard categories include non-violent crimes, specialized advice, privacy, indiscriminate weapons, suicide self-harm, and elections misinformation. The data was synthesized using few-shot prompting with real examples from the NVIDIA AEGIS AI Content Safety Dataset 2.0 as seed patterns. The dataset is stored in JSONL format, with each sample containing text, category, and label fields.
提供机构：
llm-semantic-router
