SimpleStories/SimpleStories

SimpleStories是一个包含超过200万条模型生成的短故事的数据集，旨在用于训练小型、可解释的语言模型。数据集包含高层次的概念注释，如主题、话题和风格，并预计算了多种NLP指标以辅助过滤。它是多语言的，目前有英语和日语版本，由2024年的模型生成。数据集还经过安全审查，排除了有害或冒犯性内容。README还提到了基于该数据集训练的一系列模型，并提供了学术引用的信息。

SimpleStories is a dataset of >2 million model-generated short stories. It was made to train small, interpretable language models on it. The dataset features annotations with high-level concepts such as theme, topic, and style, and includes various NLP metrics pre-computed for filtering. It is multilingual, with versions available in English and Japanese, and is generated by 2024 models. The dataset also includes safety vetting to exclude harmful or offensive content. The README also mentions a family of models trained on this dataset and provides citation information for academic use.