Wikipedia Comparable Corpora

该数据集包含了英文和德文的维基百科文章，这些文章经过对齐处理，用于训练和评估M3L-Contrast模型。此外，数据集还包括了对齐的图片，专门用于评估M3L-Contrast模型的表现。其规模包括20,000个训练元组和1,000个测试元组，任务涉及多语言和多模态的主题建模。

This dataset comprises aligned English and German Wikipedia articles, intended for training and evaluating the M3L-Contrast model. Furthermore, the dataset includes aligned images specifically used to assess the performance of the M3L-Contrast model. It consists of 20,000 training tuples and 1,000 test tuples, with its tasks involving multilingual and multimodal topic modeling.