TNauen/ImageNet-Caption-Encodings

该数据集名为ImageNet Captions and Caption Embeddings，用于图像分类和图像特征提取任务，包含英语的标注文本和文本嵌入。数据集规模在100万到1000万之间，具体用于论文《TextTeacher: What Can Language Teach About Images?》中，旨在探索语言对图像理解的作用。它提供了ImageNet图像的描述（即标注）和对应的编码表示，以支持多模态学习研究。

This dataset, named ImageNet Captions and Caption Embeddings, is designed for image classification and image feature extraction tasks, containing English captions and text embeddings. It has a size between 1 million and 10 million entries and is used in the paper TextTeacher: What Can Language Teach About Images? to explore how language can inform image understanding. It provides captions and corresponding encodings for ImageNet images, supporting multimodal learning research.