In this exercise, you will set up a Data Science Virtual Machine (DSVM) in Azure, and import the code from this GitHub repo into its Jupyterhub environment. You'll then use a Python notebook to generate and explore some image data.
- In the Azure portal, create a new Data Science Virtual Machine for Linux (Ubuntu) resource with the following settings:
- Resource Group: Create a new resource group
- Virtual Machine Name: Any unique name
- Region: Any available region
- Virtual Machine Image: Data Science Virtual Machine for Linux (Ubuntu)
- Size: NC6 Standard (GPU Family)
- Authentication type: Password
- Username: (Specify a lowercase user name of your choice)
- Password: (Specify a complex password)
- OS Disk Type: Standard SSD
- View the properties of your DSVM and determine its IP address.
-
In a new browser tab, navigate to Jupyterhub on your DSVM at https://YOUR.DSVM.IP.ADDRESS:8000. You will need to click through the certificate validation warning messages that are displayed.
-
Sign into Jupyterhub using the user name and password you specified when creating the DSVM. If the home page doesn't open after a few seconds, click the jupyter logo to open it.
-
Note that the DSVM already contains notebooks for you to explore. Then on the New menu, click Terminal and click through the certificate warnings once again until you open a terminal window.
-
In the terminal window, enter the following command to change the current directory to the notebooks folder:
cd notebooks
-
Now enter the following command to clone this GitHub repository to this folder:
git clone https://github.com/GraemeMalcolm/AI-Airlift
-
After the repo has been downloaded, switch back to the tab containing the folder tree, refresh the view if necessary, and verify that the AI-Airlift folder has been downloaded.
-
Open the AI-Airlift folder and note that it contains the files and folders from this repo.
- In the AI-Airlift/code folder, open the 01 - Exploring Images.ipynb notebook.
- Read the notes in the notebook, running each code cell and reviewing the output. Take the time to examine the code and ensure you understand what it is doing.