Let me break down clear answers to both of your questions:

1. Can I shut down my local computer while training a model on Google Cloud? #

Absolutely! Training tasks on Google Cloud run entirely on cloud-based VM instances or dedicated training services like Vertex AI Training—these resources are fully independent of your local machine. Once you’ve started the training and the cloud instance is active, the task will keep running smoothly even if you shut down your local computer, disconnect from the internet, or close any local terminals/browsers.

A quick tip: If you launched the training via an SSH connection from your local terminal, use tools like nohup, tmux, or screen to run the process in the background. This prevents the training from stopping if your SSH session accidentally drops. If you started the task directly through the Google Cloud Console, you don’t need to worry about this at all—Google’s infrastructure will handle the task lifecycle automatically.

2. Do I need to keep my local computer on when training via Jupyter Notebook on a Google Cloud instance? #

You don’t have to keep your local computer powered on, but there’s one key detail to note: If you’re running training code directly in a Notebook cell in your browser (without offloading it to a cloud background service), closing the browser or shutting down your local machine will interrupt the running cell.

To avoid this, try either of these solutions:

• Use the built-in background job features in your cloud Notebook environment (like Vertex AI Workbench), which lets you submit training as a separate cloud job that runs independently of your browser session.
• If you’re using a self-hosted Jupyter instance, run your training script via the command line with nohup or use tmux to preserve the session. This way, the training continues on the cloud VM even if you close your browser.

内容的提问来源于stack exchange，提问作者Huzo