Is there any way to use the fallback policy with the supervised response selector, similar to how it’s used with a normal intent classifier? Would be useful if the chatbot is not confident about which FAQ is being asked. If not, are there any plans to develop such a feature soon?
3 Likes