Deploy Rasa Core separately

[r4sn4 (Rasna Tomar)]

Hi All,

Is it possible to deploy Rasa Core separately like Rasa NLU?

I dont want to pass NLU model path to rasa core.

I want to pass my conversations to Rasa NLU, then pass result from NLU to Core API.

[Ghostvv (Vladimir Vlasov)]

Rasa NLU and core are now merged into one repo

[r4sn4 (Rasna Tomar)]

I want to use to only use RASA core for my usecase not RASA NLU. Is it possible?

[Ghostvv (Vladimir Vlasov)]

Rasa NLU will be there, but you don’t need to use it

[r4sn4 (Rasna Tomar)]

Rasa core accepts message in this form - {“message”: “hi”, “sender_id”: “1”}

But I want to send predicted intent from my NLU model to rasa core api directly. How can I achieve this?

[r4sn4 (Rasna Tomar)]

I checked two APIs just now -

Add a message to a tracker - http://localhost:5005/conversations/{conversation_id}/messages

Predict the next action - http://localhost:5005/conversations/{conversation_id}/predict

I think these two API will help in achieving the functionality that I want.

Is conversation_id = sender_id ?

[erohmensing (Ella Rohm-Ensing)]

@r4sn4 yes, conversation_id == sender_id Also you can always send intent format from your NLU to rasa core:

{“message”: "/intent{"entity": "some_value"}, “sender_id”: “1”}

[r4sn4 (Rasna Tomar)]

Thanks @erohmensing