locuslab/multi_password_eval

MultiPasswordEval是一个多轮密码评估数据集，它是PasswordEval数据集的多轮版本。数据集包含对话ID、系统提示、用户提示、合规和不合规的用户提示内容及其角色、密码序列、密码数量、敏感信息、正确响应和错误响应。测试集包含900个示例，数据集总大小为1,602,731字节。MultiPasswordEval is a multi-turn password evaluation dataset, which is an extension of the PasswordEval dataset. It includes conversation IDs, system prompts, user prompts, compliant and non-compliant user prompt contents and their roles, a sequence of passwords, number of passwords, sensitive information, correct responses, and incorrect responses. The test set contains 900 examples, and the total size of the dataset is 1,602,731 bytes.