MaziyarPanahi/open-perfectblend-fixed

该数据集是对原始数据集进行修复后的版本，原始数据集在Axolotl中出现了两种不同的错误。修复过程中，过滤掉了有问题的条目，确保每条对话都是有效的。过滤条件包括：对话必须是列表形式，每条消息必须是字典形式，且必须包含from和value字段，对话中必须包含至少一条来自human和gpt的消息，并且消息总数必须大于等于2。数据集包含1358785个训练样本，下载大小为1455283492字节，数据集大小为2822341964.797401字节。

This dataset contains conversational data, where each conversation consists of multiple messages, each containing a sender and content. The dataset is split into a training set with 1,358,785 samples, totaling 2.82GB in size. The download size of the dataset is 1.46GB. The original dataset had issues in Axolotl, and the author has filtered out problematic entries to fix the dataset.