in working on an experiment on a ML technique that required me to use a better machine for computational purposes, so they gave me an SSH connection to the machine. Also the data were stored in that server.
My workflow was this:
(I'm working on a headless server)
- Connect my local machine via ssh and run the script for the experiments...
On that machine I could only use vim without all my setup
If I want to change something I have to change it in my local then push the changes
I pull the changes on the remote server and then I try a new experiment.
Occasionally I had to push from the remote server the results (plots and more) and then pull them from local to work on that and push again eventually.
I think there is a flaw in this, and there's a better way to manage all of these things.
Do you have some ideas?
What i need is just a clever way to do not push every change i do.
CodePudding user response：
Another alternative is to use an IDE like VSCode with the Remote - SSH extension, following this tutorial.
That way, your local VSCode, on your local machine, displays and edits directly files on the remote machine, without you having to pull/push them.
Depending on that extension, you might still need a separate SSH session in order to git add/commit those modified files.