Exercise Part 2: Creating Compute Resources in Azure Machine Learning
Now that you have your Azure Machine Learning workspace, it’s time to set up the compute resources. These are the powerhouse behind your machine learning models, where all the training and processing magic happens.
Step 1: Navigating to the Compute Page
- Finding Your Tools: Head over to the Compute page in the Azure Machine Learning studio. This is your control center for managing compute targets, the engines of your data science escapades.
Step 2: Understanding Compute Targets
- Types of Compute Resources: There are four main types of compute resources you can create:
- Compute Instances: Think of these as personal data science workstations for playing with data and models.
- Compute Clusters: These are scalable clusters of virtual machines, ready to tackle your experiment codes.
- Inference Clusters: The stage where your trained models perform their predictive magic.
- Attached Compute: This lets you link up existing Azure compute resources, like Virtual Machines or Azure Databricks clusters.
Step 3: Creating a Compute Instance
- Setting Up Your Workstation:
- Virtual Machine Type: Choose CPU.
- Virtual Machine Size: Select Standard_DS11_v2 (you can find this under ‘Select from all options’).
- Compute Name: Give your instance a unique identifier.
- Enable SSH Access: Keep this unselected.
Step 4: Setting Up a Compute Cluster
- For Training Your Model:
- Virtual Machine Priority: Set to Dedicated.
- Virtual Machine Type: Again, choose CPU.
- Virtual Machine Size: Stick with Standard_DS11_v2.
- Compute Name: Name this one uniquely too.
- Node Settings: Set the minimum number of nodes to 0 and the maximum to 2.
- Idle Time: Set ‘Idle seconds before scale down’ to 120 seconds.
- Enable SSH Access: Keep this unselected as well.
Important Note:
- If you’re just here for a visit and not completing this module, remember to stop your compute instance. This prevents any unexpected charges on your Azure subscription. Keep in mind that setting up these compute targets takes some time. Feel free to jump to the next unit while they’re getting ready.
And that’s it! You’re now all set to start training models and exploring data on your very own Azure Machine Learning playground. Happy computing!
Leave a Reply