jan-hq/sailor-instruct-merge-short

该数据集包含了对话信息，每个示例包括提示ID、消息（包含角色和内容）、语言、类别、映射类别、各种分数和排名信息，以及用户消息的token长度。数据集分为训练集，并提供了相关的文件路径配置。

The dataset consists of dialogue information, including prompt ID, messages (with role and content), language, category, mapped category, various scores and ranking information, as well as the token lengths of user messages. The dataset is split into a training set and provides related configuration for data file paths.