lhoestq/tmp-lmsys

该数据集包含对话信息，每个对话有唯一的会话ID，并标注了使用的模型。对话内容被分为发言内容和角色，同时记录了对话的轮次和使用的语言。数据集还包含了对对话内容进行分类的openai_moderation信息，包括是否涉及骚扰、仇恨、自伤、性内容、暴力等分类，并为每个分类提供了详细分数和是否标记的标识。数据集分为训练集，并提供了训练集的示例数量和大小信息。

The dataset includes conversation information, each with a unique session ID and the model used is labeled. The conversation content is divided into speech content and role, and the turn of the conversation and the language used are recorded. The dataset also contains openai_moderation information that classifies the conversation content, including whether it involves harassment, hate, self-harm, sexual content, violence, etc., and provides detailed scores and flags for each classification. The dataset is divided into training sets and provides information on the number of examples and size of the training set.