Toxic Comments Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/ejones313/auditing-llms/tree/main/data
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含按标记数量分类的有毒评论，具体包含68,332条只有一个标记的有毒评论，592条有两个标记的有毒评论，以及三个标记的有毒评论。该数据集的任务是对文本进行分类并与图像进行对齐。
This dataset comprises toxic comments categorized based on the number of annotation labels. Specifically, it includes 68,332 toxic comments with only one annotation label, 592 toxic comments with two annotation labels, and toxic comments with three annotation labels. The task of this dataset is to perform text classification and align text with corresponding images.
