jackzhang/BeaverTails-promptembed

该数据集包含多个字段，主要用于分类和评估文本内容的安全性。数据集中的每个条目包含一个提示（prompt）和对应的响应（response），以及多个类别标签（category），这些标签用于标识内容是否涉及特定类型的不当行为或敏感话题，如虐待动物、虐待儿童、歧视、仇恨言论等。此外，数据集还包含一个is_safe字段，用于标识内容是否安全，以及一个prompt_embedding字段，表示提示的嵌入向量。数据集被分为多个分割，包括330k_train、330k_test、30k_train和30k_test，每个分割都有对应的字节大小和示例数量。

This dataset contains multiple fields primarily used for classifying and evaluating the safety of text content. Each entry in the dataset includes a prompt and a corresponding response, along with multiple category labels that indicate whether the content involves specific types of inappropriate behavior or sensitive topics, such as animal abuse, child abuse, discrimination, hate speech, etc. Additionally, the dataset includes an is_safe field to indicate whether the content is safe and a prompt_embedding field representing the embedding vector of the prompt. The dataset is divided into several splits, including 330k_train, 330k_test, 30k_train, and 30k_test, each with corresponding byte sizes and example counts.