patrickamadeus/coco_caption_val_unique

该数据集是一个包含5000个训练样本的图像数据集，每个样本包括图像、标题、文件名和图像ID。图像特征的数据类型为image，标题和文件名的数据类型为string，图像ID的数据类型为int64。数据集总大小为257381003字节，下载大小为257038293字节。训练数据文件位于data/train-*路径下。

This dataset is an image dataset containing 5000 training samples, each of which includes an image, a caption, a filename, and an image ID. The image feature has a data type of image, the caption and filename have a data type of string, and the image ID has a data type of int64. The total size of the dataset is 257381003 bytes, and the download size is 257038293 bytes. The training data files are located under the path data/train-*.