I've just been using ideal conversations as training data from past campaigns and agent calls. Interested to see how the conversational sciences dataset looks like