LF-WikiSeeAlso-320K

该数据集在训练时使用了维基百科数据转储中干净的文本部分，旨在服务于问答任务。在预处理阶段，为了确保在微调任务中的公平性，该数据集排除了“参见”部分。该数据集的规模为320K，其任务是实体分类。

This dataset employs clean text segments extracted from Wikipedia data dumps during the training phase, and is developed for question answering tasks. During the preprocessing stage, the "See also" sections were excluded from this dataset to ensure fairness in downstream fine-tuning tasks. This dataset consists of 320K instances, and its target task is entity classification.