PRLM/oasst2

这是一个包含对话数据的训练集，每个样本包括对话内容、对话角色、质量评分等信息。数据集特别针对了包含特定关键词（如OpenAI、ChatGPT等）的对话进行了过滤，同时移除了包含空消息的对话。数据集经过筛选和排序，以优化质量评分较高的样本。

This is a training dataset containing dialogue data, with each sample including dialogue content, dialogue roles, quality scores, and other information. The dataset specifically filters out dialogues containing certain keywords (such as OpenAI, ChatGPT, etc.) and removes dialogues with empty messages. The dataset has been filtered and sorted to prioritize samples with higher quality scores.