So sorry for the delay here! totally lost track of this thread

Should I be deploying the entire docker file for every model, with the updated requirements?

It's for every "environment" i.e. if models need the same set of python packages , you canshare