I am currently training a network that takes several hours. When I spin up a Floydhub instance on my laptop, then close the laptop or disconnect from internet, the notebook I am running gets shutdown but the GPU session is still running, which is pretty wasteful for my credits.
Is there a way to keep my notebook running while it is not opened in my browser? I'd like to train my model even when my computer is shut down. Is this a question of dedicated vs preemptible instances?