Doubt regarding architecture of Rasa Deployment

In the following image, I believe that blue Filesystem box contains trained .tar model files stored in google drive or amazon S3 and the purple Filesystem box contains model files trained/uploaded on Rasa X UI.

I have used quick install script for deployment. If I don’t upload and connect any trained model file manually (to blue Filesystem box), does the deployment automatically consider active model file on rasa X (Purple box) as the primary model while interacting with Rasa open source using input-output channels?

Also, How do we connect the uploaded model file (on google drive or S3) to rasa open source while deploying using Quick Install Script.

Hi @gaushh storing models in S3/other remote storage is not supported with Rasa X. You need to train a model on Rasa X, or upload it manually in order to use it on there

@akelad Thanks for the response.

If models can’t be connected to remote storage, what does the blue Filesystem box represent? And, if I were to deploy my bot without Rasa X, how would I connect the pre-trained model to the deployment?

Hi @gaushh thanks for pointing this out, that blue Filesystem box shouldn’t be there at all actually. Everything that Rasa Open Source gets in terms of models comes from Rasa X. We will update this diagram in our docs soon