Luo-wj/Hard_VQA_RL_diverse

该数据集是一个多模态数据集，包含图像和文本数据。数据集的特征包括图像列表、问题字符串、答案字符串列表、类别字符串、数据源字符串和图像路径字符串。数据集分为训练集和测试集，训练集有10159个样本，测试集有2032个样本。数据集的下载大小为2340543648字节，总大小为2343052400.125字节。

This dataset is a multimodal dataset containing both image and text data. The features of the dataset include a list of images, a question string, a list of answer strings, a category string, a data source string, and an image path string. The dataset is divided into a training set and a test set, with 10159 samples in the training set and 2032 samples in the test set. The download size of the dataset is 2340543648 bytes, and the total size is 2343052400.125 bytes.