LLMPopcorn
收藏资源简介:
LLMPopcorn数据集是由英国格拉斯哥大学、中国山东大学和西班牙电信科学研究机构创建的,包含根据用户提示生成的视频标题和视频生成提示。数据集分为具体和抽象两种用户提示类型,涵盖了动漫、美食、日常生活分享、电影与电视、游戏五大类别,每个类别各有二十个提示。该数据集用于评估大型语言模型在生成流行微视频方面的性能。
The LLMPopcorn dataset was developed by the University of Glasgow in the United Kingdom, Shandong University in China, and the Telefónica Scientific Research Institute. It comprises video titles and video generation prompts generated from user prompts. The dataset is categorized into two types of user prompts: concrete and abstract, and covers five core categories, namely animation, food, daily life sharing, film and television, and gaming, with 20 prompts allocated to each category. This dataset is employed to evaluate the performance of large language models (LLMs) in generating popular micro-videos.




