shalanova/benchmark-4-arabic-m2m

该数据集是一个关于内容安全的数据集，包含多样化的不安全类别（如有害指令、敏感话题、对抗性改写等）和不一定遵循典型越狱模板的提示。数据集旨在增加多样性和分布变异性，使基于相似性的检测更具挑战性，并为跨语言迁移提供压力测试。数据集包含1,000个提示（500个安全/500个不安全），列包括原始提示文本、标签（0表示安全，1表示不安全）、阿拉伯语翻译以及阿拉伯语模型的余弦相似度分数。

This dataset is related to content safety, including diverse unsafe categories (e.g., harmful instructions, sensitive topics, adversarial rephrasings) and prompts that do not necessarily follow canonical jailbreak templates. The dataset aims to increase diversity and distributional variability, making similarity-based detection more challenging and providing a stress-test for cross-lingual transfer. The dataset contains 1,000 prompts (500 safe / 500 unsafe), with columns including the original prompt text, label (0 for safe, 1 for unsafe), Arabic translation, and cosine similarity score for the Arabic model.