GPT-4 Generated Stories
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/UBC-NLP/arastories
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了由GPT-4-Turbo API生成的3,000个故事，涵盖了三种阿拉伯语变体：现代标准阿拉伯语、埃及阿拉伯语和摩洛哥阿拉伯语。每个语种均有1,000个故事，这些故事是基于一套精心设计的提示语生成的。整个数据集规模达到了3,000个样本，任务旨在进行故事生成。
This dataset contains 3,000 stories generated via the GPT-4-Turbo API, covering three Arabic variants: Modern Standard Arabic (MSA), Egyptian Arabic, and Moroccan Arabic. Each of these variants has 1,000 stories, which were created based on a meticulously crafted set of prompts. The full dataset consists of 3,000 samples in total, with the core task being story generation.
