Before I started using GNU Screen and tmux, when I needed to train a machine learning model on a remote machine, the following describes the typical steps I had to take:
I open up a terminal on my local machine, ssh to the remote node, and run the program, which prints out the training progress. So far so good.
But as it is running, I want to monitor resource usage. So I open another terminal, ssh to the remote machine, and then run
top. Looks alright.