CAS-SIAT-XinHai/FontHalu-mitigation

该数据集用于研究多模态大语言模型中基于字体的幻觉问题，包含图像和文本数据。数据集包含多个配置(Cluster-10, Cluster-200, Same, SemAna, test)，每个配置都有训练集或测试集，特征包括图像名称、图像列表和标签等。This dataset is for studying font-based hallucinations in multimodal large language models, containing both image and text data. The dataset includes multiple configurations (Cluster-10, Cluster-200, Same, SemAna, test), each with training or test sets, and features such as image names, image lists, and labels.