Hi, can i use Rasa Nlu only to flag wrong intens and then retrain the model. Because in my bot i am not using rasa core. Thanks.
Continuing the discussion from Can i use Rasa X only for Rasa NLU?:
I imagine one might need to manually add messages to the Tracker as described in the tracker docs, as the Rasa X UI is probably only capable of presenting full conversations and not single messages and intent predictions. However the Tracker API is only available for rasa servers and not rasa nlu-only servers. Can someone elaborate more on this and whether there are any workarounds? I am also using the NLU only as a rule based dialog policy is better suited for my use case.
I would also love to see some ideas or workaround for this. More concretely, I am trying to see if Rasa X can be of any use to quickly label/annotate or correct the existing intent only, let’s say going through line by line of a list of sentences that a bunch of users wrote. Each sentence is taken from different conversations.
Bit late to the party, but see these answers