KazCulture is a culturally focused Kazakh dataset consisting of 16,137 human-crafted passage–question–answer (PQA) triplets designed to evaluate and enhance large language models’ understanding of Kazakh cultural knowledge. The passages were collected from 11 authoritative books on Kazakh culture and the Koshpendiler.kz website, covering topics such as customs and traditions, traditional music and instruments, beliefs and… See the full description on the dataset page: https://huggingface.co/datasets/issai/KazCulture.
