Twitter Dialog Corpus

该数据集是从推特上收集的，包含了2.6百万条（消息，回复）对，经过特定长度限制的筛选后，保留了2百万条样本。此外，数据集还根据历史消息和回复的长度限制进行了进一步筛选。规模达到了2百万样本，适用于开放域对话生成任务。

This dataset is collected from Twitter, initially containing 2.6 million (message, reply) pairs. After filtering based on specific length constraints, 2 million samples are retained. Furthermore, the dataset undergoes additional filtering according to the length limits of historical messages and their corresponding replies. With a total of 2 million samples, this dataset is suitable for open-domain dialogue generation tasks.