mlfoundations-dev/stackoverflow_100000_samples

这是一个包含指令和完成情况的对话数据集，每个样本包括一个指令、一个完成情况、原始行索引以及一个包含对话来源和内容的列表。数据集仅包含一个训练集，共有100000个样本。

This is a dialogue dataset containing instructions and completions. Each sample includes an instruction, a completion, an original row index, and a list containing the source and content of the conversation. The dataset consists of only one training set with a total of 100,000 samples.