CC12M
收藏arXiv2025-09-30 收录
下载链接：
https://arxiv.org/pdf/2102.08981v1.pdf
下载链接
链接失效反馈官方服务：
资源简介：
该数据集用于训练基于场景的转换器模型，包含了文本与图像相对应的对。此外，该数据集与其他数据集如CC和MS-COCO结合使用，其规模达到了3500万文本图像对，主要用于文本到图像生成的任务。
This dataset is developed for training scene-based Transformer models and includes paired text and image samples. Additionally, when utilized in combination with other datasets such as CC and MS-COCO, the total scale of the combined dataset reaches 35 million text-image pairs, and it is mainly applied to text-to-image generation tasks.
