Creating Vertex AI Notebook & JupiterLab Setup

• In the Google Cloud Console, first check if you are in the right project where you want to create a Vertex AI Workbench.

• In the search service section search for Vertex AI , and in Vertex AI go to Workbench on the left panel.

• Verify that you are on the USER_MANAGED NOTEBOOKS tab.

• Click on “NEW NOTEBOOK” on the top > TensorFlow Enterprise > Select the Version & with/ without GPU based on requirement.

• In the Pop-up select advanced settings > give a name to the notebook & check the region/zone where the notebook instance will be launched(Notebook is a VM instance under the hood).

• Once the notebook is created, click on the “OPEN JUPYTERLAB”  link in the list view, this will open the Jupyter notebook in the new browser tab.

• On the Launcher page, double click on Terminal to configure the Gitlab SSH & clone the repository.

• In the terminal enter the command “ssh-keygen” , to generate a key(which will be linked with the Gitlab account)

• After key is generated, check the tree structure to see if id_rsa.pub file is generated by using the “tree .ssh” command

• Put the “cat id_rsa.pub” command to open the file and view the SSH Key content.

• Copy the whole content & place it in the SSH KEY settings in the gitlab account , and click on Add key to complete the SSH key setup. Note that this is a one time setup, after which one will be able to clone the gitlab project/s to the notebook.Add to the Key text box in the Gitlab.

• Copy the Clone URL from gitlab for the project to be cloned

• Back in the Jupyter notebook enter the command “git clone+ the url copied from gitlab”.

• You will see that the project folder is visible in the JupyterLab notebook.