How do we want to pull trained models?

[bethac07]

Right now, DDP assumes that the trained model you want to use is on your machine in the correct location, due to the Docker container.

I suspect we'd rather have this as a configurable parameter - @jccaicedo , what do you think?