Document Understand ML On Premise Clarifications

Hello Everyone,

I have a question regarding the on premise set-up, on the side of ML extractor. The documents are not sent to the cloud since it is an on premise set-up? Who hosted the ml skill or pre-trained model (DU Endpoint), is it the AI Fabric server? So the documents are sent to the AI Fabric server for processing? is my understanding correct? thanks for your feedbacks

if you’ve set up Orchestrator & AIFabric on premise, the ML Skills are hosted on the AIFabric server.

We have an on-prem Orchestrator and AI Fabric install. Were you able to connect to AI Fabric and select a skill in studio? OUr skills list is empty.