mlfoundations-dev/stackoverflow_10000tasks_.75p

这是一个包含对话和指令的数据集，共有50000个训练样本。数据集的字段包括指令、完成情况、对话（包括发送者和信息内容）、种子任务、生成的命令提示、生成的命令、命令响应、最终指令和最终响应。

This dataset contains dialogues and instructions, with a total of 50,000 training samples. The fields of the dataset include instructions, completion status, conversations (including sender and message content), seed tasks, generated command prompts, generated commands, command responses, final instructions, and final responses.