2 datasets found

The Building a conversational agent overnight with dialogue self-play dataset is a benchmark for conversational AI.

The dataset is constructed using a dialogue self-play approach, where synthetic dialogues are generated to grant that phenomena uninvestigated are present only at test time and...