govtech/lionguard-2-synthetic-instruct

LionGuard 2数据集是一个新加坡语境下的多语言内容审查分类器子集，包含用聊天机器人风格重写的论坛评论，以及与文本相关的审查标签，如总体安全性、仇恨、侮辱等，使用0/1/2的标签体系来表示内容的危险程度。LionGuard 2 is a subset of a multilingual content moderation classifier for the Singapore context, including chatbot-style rewritten forum comments paired with moderation labels such as overall safety, hate, insults, etc., using a 0/1/2 label system to indicate the risk level of the content.