CMU DoG
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/festvox/datasets-cmu_dog
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为CMU文档扎根对话数据集，包含大约4,000个对话，覆盖了90多个话题。其中包含74,485个三元组信息，涉及14,689个实体和42种关系。该数据集的规模大约为4,000个对话，任务定位于知识扎根的对话生成。
This dataset is named CMU Document-Grounded Dialogue Dataset. It consists of approximately 4,000 conversations spanning over 90 topics, and contains 74,485 knowledge triples involving 14,689 entities and 42 distinct relationship types. With a scale of around 4,000 conversations, this dataset is targeted at knowledge-grounded dialogue generation.
