Hi community
I’m struggling to understand how Rasa X and Rasa NLU/Core work together on production.
I set up Rasa X on my Ubuntu server using the docker image. That worked all fine. Now I would like to start designing my bot and connecting to it with my application. I struggle to figure out the right approach. My two different ideas:
Approach A - I only install Rasa X (which contains a production Rasa instance) and not another Rasa instance. I send the message-request from my client directly to Rasa X. In Rasa X I define which model is the production model.
If this is the right approach:
- Where do I access the configuration files (domain.yml, stories.md, config.yml) and how do I adjusted them manually (for example to change the language and to put these files into version control)?
Approach B - I install Rasa X and setup an independent Rasa NLU/Core docker instance. In the Rasa X documentation it is written about connection, importing and migrating Rasa NLU/Core - three different approaches - this confuses me.
If this is the right approach:
- What is the right workflow? I setup Rasa and define the config.yml. I train the other files in Rasa X and then copy the model back to my Rasa instance?
I would really appreciate to understand the right workflow.