CTCT-CT2/ChangeMore-prompt-injection-eval

ChangeMore-prompt-injection-eval数据集是为了支持大型语言模型（LLM）的提示注入检测能力评估而设计的。该数据集通过一种系统性的数据生成算法自动产生大量高质量的攻击样本，显著丰富了中文语境下LLM的安全评估生态系统。数据集是ChangeMore数据集的关键组成部分，专注于各种攻击类型和上下文环境中的提示注入攻击场景。它为增强中文LLM在提示注入防御方面的训练和评估，以及研究新兴的攻击模式和防御机制提供了强有力的支持。

The ChangeMore-prompt-injection-eval dataset is designed to support the evaluation of prompt injection detection capabilities in large language models (LLMs). To address the lack of Chinese-language prompt injection attack samples, a systematic data generation algorithm has been developed to automatically produce a large volume of high-quality attack samples, significantly enriching the security evaluation ecosystem for LLMs, especially in the Chinese context. The dataset is a key component of the ChangeMore Dataset, focusing on prompt injection attack scenarios across diverse attack types and contextual settings. It provides strong support for enhancing the training and evaluation of Chinese LLMs in prompt injection defense and enabling research into emerging attack patterns and defense mechanisms.