RedEval

该数据集是为了对基于Twitter数据集训练的模型进行域外评估而构建的，其数据来源于Reddit。它旨在补充MMSD2.0数据集，用于域外评估。该数据集包含了395个来自“sarcasm”子版块的讽刺示例和609个来自其他子版块的负面示例，其任务是进行多模态讽刺检测。

This dataset was developed for out-of-domain evaluation of models trained on Twitter-based datasets, with its data originating from Reddit. It is designed to supplement the MMSD2.0 dataset for out-of-domain evaluation tasks. The dataset comprises 395 sarcasm examples from the "sarcasm" subreddit and 609 negative examples from other subreddits, targeting the task of multimodal sarcasm detection.