Qor´gau
收藏资源简介:
Qor´gau是一个专门为评估哈萨克语和俄语大型语言模型安全性而设计的 novel 数据集。该数据集由阿布扎比人工智能研究学院创建,包含针对哈萨克斯坦独特的双语环境定制的区域特定和文化相关提示,共有8169条问题,涵盖六个风险领域和17种危害类型。数据集的创建过程包括翻译和本地化、创建区域特定问题以及收集代码切换提示。该数据集旨在解决在哈萨克斯坦等双语地区对大型语言模型安全性的评估问题。
Qor´gau is a novel dataset specifically designed to evaluate the safety of large language models (LLMs) for Kazakh and Russian languages. It was developed by the Abu Dhabi Artificial Intelligence Research Institute. The dataset includes region-specific and culturally relevant prompts tailored to Kazakhstan's unique bilingual environment, with a total of 8169 questions covering six risk domains and 17 hazard categories. The dataset's construction process involves translation and localization, creation of region-specific questions, and collection of code-switching prompts. This dataset aims to address the challenge of evaluating LLM safety in bilingual regions such as Kazakhstan.




