ziyu3141/rich_feedback_train_with_image

该数据集包含多个特征，如文件名、美学评分、伪影评分、错位评分、总体评分、伪影热图、错位热图、错位标记标签、提示词和图像等。数据集分为训练集，包含15810个示例，总大小为101068478704字节。下载大小为1715550658字节。

This dataset is primarily used for image quality assessment, containing various features related to image quality such as aesthetics score, artifact score, misalignment score, and overall score. Additionally, the dataset includes artifact heatmaps and misalignment heatmaps, as well as misalignment token labels and prompt information. The image data is stored in binary format. The dataset is divided into a training set, containing 15810 samples, with a total dataset size of 101068478704 bytes and a download size of 1715550658 bytes.