celeb-1000

该数据集包含了1000位名人的图片及相关信息，主要用于视觉-语言任务中训练模型。为了模拟遗忘场景，数据集中已过滤掉已知实体，从而支持在保护隐私方面对遗忘学习方法有效性的实验研究。该数据集的任务是测试大型视觉-语言模型中的遗忘学习方法。

This dataset contains images and associated information of 1000 celebrities, and is primarily used for training models in vision-language tasks. To simulate forgetting scenarios, known entities have been filtered out from the dataset, thereby supporting experimental studies on the effectiveness of forgetting learning methods for privacy preservation. The task of this dataset is to test forgetting learning methods in large vision-language models.