GPT-Neo 2.7B

该数据集是一个大型语言模型，旨在生成用于评估文本到图像模型漏洞的提示。它采用上下文学习的方式来生成多样化的对抗性提示，其任务是对文本到文本模型进行红队攻击演练。This dataset is a large language model (LLM) designed to generate prompts for evaluating the vulnerabilities of text-to-image models. It employs in-context learning to produce diverse adversarial prompts, with its core task being to conduct red teaming exercises against text-to-text models.