tomg-group-umd/fictionalqa

FictionalQA数据集是一个专门用于研究事实记忆和逐字序列记忆双重过程的合成数据集。它由关于虚构事件和各种相关事实的合成文本组成，以及关于这些事实的问题和答案对。数据集支持各种语言模型训练实验，并提供结构化的文档和问答对，适合作为预训练和微调实验的数据。

The FictionalQA dataset is a synthetic dataset designed to study the dual processes of fact memorization and verbatim sequence memorization. It consists of synthetically generated text about fictional events and related facts, along with question-answer pairs about these facts. The dataset supports various language model training experiments and provides structured documents and question-answer pairs suitable for pretraining and fine-tuning tasks.