I just wanted to mention that the training data after interactive learning aren’t separated based on retrieval intents but are just stacked all together below for example intent: chitchat.
As I see from Response retrieval models - Interactive learning, Response Retrieval isn’t integrated in interactive learning, just a request for future
Actually, if it’s an open issue, I would like to see if I can contribute.
Hey @magda
As Daksh mentioned in the post you’ve linked, that’s still an experimental feature, so we don’t have concrete plans for it just yet, but please stay tuned!
1 Like