Hi!
Currently, when you have multiple utterances examples for an utterance, one of them is randomly picked. In addition, slots that are not set are simply filled with None
.
Is there a way to override the mechanism that selects which utterance to use? For example, we could have some utterance examples that contain a slot and others that do not have it. We could pick the right utterance depending on which slots have the information filled.
utter_greet:
- text: Hello {name}!
- text: Hey there
Right now, we rely on an external NLG server to do this (which works fine) but it would be great to have this option without using an NLG server.
Cheers, Nicolas