Weibo Benchmark Dataset

该数据集是一个拥有超过4亿个训练对的开放领域中文对话数据集，旨在评估对话生成模型的性能。它不仅包含了用于评估的自动评价指标，还包含了针对回复质量的人工评估。数据集的规模达到了4000多万训练对和3200多测试对，其核心任务是对话生成。

This is an open-domain Chinese dialogue dataset containing over 400 million training pairs, which is designed to evaluate the performance of dialogue generation models. It not only includes automatic evaluation metrics for model assessment, but also provides human evaluations of response quality. The dataset comprises over 40 million training pairs and more than 3,200 test pairs, and its core task is dialogue generation.