HiXSTest
收藏arXiv2025-09-30 收录
数据链接：
官方服务：
资源简介：
该数据集旨在评估大型语言模型在不同文化背景下的过度安全声明，它是评估大型语言模型综合工具包的一部分。具体任务是对大型语言模型的过度安全性进行评估。
This dataset is designed to evaluate overly safe statements generated by large language models (LLMs) across diverse cultural backgrounds. It is part of a comprehensive toolkit for large language model evaluation, and its specific task is to assess the over-safety of LLMs.
提供机构：
WalledEval
