difference between "notebooks" and non-notebooks DLVM images

[Jason Brancazio]

Hi DL Platform team,

When launching a Vertex AI Workbench instance with gcloud workbench instances create, I can specify a --vm-image-name from the deeplearning-platform-release project.

The following are image names for PyTorch 2.2 with CUDA 12.1. What is the difference between the "notebooks" and non-notebooks images? Which is a better choice for Vertex AI Workbench instances, and why?

pytorch-2-2-cu121-notebooks-v20240514
pytorch-2-2-cu121-notebooks-v20240514-debian-11
pytorch-2-2-cu121-notebooks-v20240514-debian-11-py310
pytorch-2-2-cu121-notebooks-v20240514-py310
pytorch-2-2-cu121-notebooks-v20240514-ubuntu-2004
pytorch-2-2-cu121-notebooks-v20240514-ubuntu-2004-py310
pytorch-2-2-cu121-notebooks-v20240514-ubuntu-2204
pytorch-2-2-cu121-notebooks-v20240514-ubuntu-2204-py310


pytorch-2-2-cu121-v20240514
pytorch-2-2-cu121-v20240514-debian-11
pytorch-2-2-cu121-v20240514-debian-11-py310
pytorch-2-2-cu121-v20240514-py310
pytorch-2-2-cu121-v20240514-ubuntu-2004
pytorch-2-2-cu121-v20240514-ubuntu-2004-py310
pytorch-2-2-cu121-v20240514-ubuntu-2204
pytorch-2-2-cu121-v20240514-ubuntu-2204-py310