vikhyatk/st-vqa

该数据集包含图像和问答对（QAs）两个主要特征。图像特征的数据类型为图像，问答对特征包含问题和答案列表，问题为字符串类型，答案为字符串列表。数据集仅包含一个训练集分割，共有18921个样本，总大小为2951305494.152字节，下载大小为2995367927字节。数据文件位于data/train-*路径下。

The dataset contains two main features: image and qas. The image feature is of image data type, and the qas feature is a list containing question and answers sub-features, where question is of string type and answers is a list of strings. The dataset includes only a training split with 18,921 samples, totaling 2,951,305,494.152 bytes in size, and a download size of 2,995,367,927 bytes. The data files are located under the data/train-* path.