ShreyashDhoot/specshield

这是一个包含文本和图像数据的训练集，其中文本数据包括干净提示语和对抗性提示语，图像数据包括原始图像、被攻击的图像、遮罩图像和修复后的图像。每个样本还包含了模型信息以及图像的质量评分、真实感评分和准确性评分。

This is a training dataset containing text and image data, including clean prompts and adversarial prompts, and image data including original images, attacked images, masked images, and inpainted images. Each sample also includes model information and image quality scores, realism scores, and accuracy scores.