Home - Azure-Samples/locutus GitHub Wiki

This tutorial is designed to get you up and running with the end-to-end sample show at Microsoft BUILD 2022. This section is devoted to setting up your Azure Machine Learning Workspace

Prerequisites

Generally, it is easier to run these exercises in the cloud (given that part of the exercise is creating custom environments). If you want to run these things locally you need to have either a virtual or conda environment that supports PyTorch.

Other requirements are:

• An Azure subscription that you have the Owner Role to create resources.
• PowerShell v5.0+
• Azure Command-Line Interface (az-cli) v2.36.0+

Setup

Setup has been greatly simplified by the inclusion of bicep templates that will correctly set up your AzureML Workspace and create two compute types needed for the other portions of the sample.

The following steps will accomplish this task:

1. Clone the repo
2. Login to your azure account using the Azure CLI (this will open a browser and ask for credentials - the output will be the list of subscriptions available in your account):

az login

3. Select your default subscription (from the output subscription list from above):

az account set --subscription <YOUR_SUB_GUID>

4. Run this PowerShell Command with the desired name of your app (something unique) and the desired Azure Region

./provision.ps1 -name <YOUR_APP_NAME> -location <LOCATION|i.e. westus2>

If you get an error saying that a resource deployment operation failed because you don't have enough quota for the Standard NC family of compute, you can follow the instructions in the docs to increase your quota.

This runs a simple PowerShell script that calls an az CLI command to provision the resources needed to create and run and AzureML workspace. The process should look something like this while it is running:

Successful completion of the process looks like this:

Workspace

Once this is done you will have a brand new AzureML workspace where you can can complete any of the subsequent tasks. To get started head over to ml.azure.com and open your workspace.

There should be two alerts in your newly created workspace: these correspond to the successful creation of your compute environments needed to accomplish the other tasks.

In my case, these compute environments are called compute-6lff and cluster-6lff. These names will be similar to the compute names created in your environment and can be accesses by clicking on the "Compute" link. There are two tabs there that will show you the names of your Compute Instance (for working with Notebooks) and your Compute Cluster (for running longer ML Pipelines):

Summary

In this exercise you created and AzureML Workspace and associated compute environments. Feel free to look around or head over to another task which interests you!