Helder5788/truth_hyperplane

该数据集包含图像及其相关描述，每个图像可能有多个描述，其中包括一个或多个描述可能存在矛盾。数据集字段包括图像ID、数据集来源、图像本身、描述、矛盾描述、是否被矛盾描述所选中的布尔值、矛盾描述的类型以及不同模型生成的描述。数据集分为测试集，包含一定数量的图像和描述示例。

The dataset consists of images and their associated captions, each image may have multiple captions, some of which may be contradictory. The dataset fields include image ID, dataset source, the image itself, captions, contradictory captions, a boolean indicating whether the caption is selected as contradictory, the type of contradiction, and captions generated by different models. The dataset is split into a test set, containing a certain number of image and caption examples.