UCSC-VLAA/gpt-edit-simpler

该数据集包含了文本编辑任务的图像输入和输出，以及相关的说明和描述。数据集中的每个样本都有一个唯一的标识符，并且标注了数据来源（如hqedit、omniedit、ultraedit）。此外，每个样本还包含了任务类型、输入图像、输出图像、说明文本、说明类型（如原始、重写、复杂）、输入描述以及原始说明文本。数据集被划分为训练集，其中包含了大量的样本。

The dataset consists of image inputs and outputs for text editing tasks, along with related instructions and descriptions. Each sample in the dataset has a unique identifier and is labeled with the data source (such as hqedit, omniedit, ultraedit). Additionally, each sample includes the task type, input image, output image, instruction text, instruction type (such as original, rewrite, complex), input description, and original instruction text. The dataset is split into a training set, which contains a large number of samples.