AllenNella/marvl

MaRVL（多语言视觉推理数据集）是一个包含多种语言（包括id、sw、ta、tr、zh、en）的视觉推理数据集。数据集包含四种特征：id、图片、假设和答案。图片是一种视觉信息，假设和答案是文本信息。数据集根据语言被划分为六个部分，每个部分包含不同数量的样本。该数据集可用于跨语言和文化背景的视觉推理任务。

MaRVL (Multilingual Visual Reasoning Dataset) is a visual reasoning dataset that includes multiple languages (id, sw, ta, tr, zh, en). The dataset contains four features: id, image, hypothesis, and answer. Image is a visual information, hypothesis and answer are text information. The dataset is divided into six parts according to languages, each part containing a different number of samples. The dataset can be used for visual reasoning tasks across different languages and cultural backgrounds.