cultural_evaluation-kalahi
收藏资源简介:
Kalahi数据集旨在评估大型语言模型(LLMs)在生成与菲律宾文化相关的响应方面的能力,特别是在共享知识和伦理方面。该数据集包含一个与多项选择题(MCQ)兼容的版本,专门用于评估菲律宾文化的表示。数据集仅包含Tagalog(tl)语言的分割,共有150个样本,统计信息包括不同模型(GPT-4o、Gemma 2、Llama 3)的token数量。数据集的许可证为CC BY 4.0,适用于评估任务和SEA-HELM排行榜。
The Kalahi dataset is designed to evaluate the capability of Large Language Models (LLMs) to generate responses related to Philippine culture, particularly in the context of shared knowledge and ethics. This dataset includes a version compatible with Multiple-Choice Questions (MCQ) specifically for assessing the representation of Philippine culture. The dataset only has splits in Tagalog (tl), with a total of 150 samples, and its statistics cover the token counts of different models including GPT-4o, Gemma 2 and Llama 3. The dataset is licensed under CC BY 4.0, and is applicable to evaluation tasks and the SEA-HELM leaderboard.




