Comment by Diris

Very cool!!! I had this idea a while. Is the conversational part of the dataset open?