Working with Compute Contexts in Azure Machine Learning

Question 1

In Azure Machine Learning, data scientists can run experiments based on scripts that process data, train machine learning models, and perform other data science tasks. The runtime context for each experiment run consists of two elements. Which ones?

Answer 1

The environment for the script, which includes all packages on which the script depends.

The compute target on which the environment will be deployed and the script run.

Question 2

What are ways to create environments in Azure Machine Learning

Answer 2

Creating an environment from a specification file

Creating an environment from an existing Conda environment

Creating an environment by specifying packages

Question 3

What are compute targets

Answer 3

Compute targets are physical or virtual computers on which experiments are run

Question 4

The ability to assign experiment runs to specific compute targets helps you implement a flexible data science ecosystem in some ways. Which ones

Answer 4

Code can be developed and tested on local or low-cost compute, and the moved to more scalable compute for production workloads

You can run individual processes on the compute target that best fits its needs. For example, by using GPU-based compute to train deep learning models, and switching to lower-cost CPU-only compute to test and register the trained model

Question 5

One of the core benefits of cloud computing is the ability to manage costs by paying only for what you use. In Azure Machine Learning, you can take advantage of this principle by defining compute targets that can do what?

Answer 5

Start on-demand and stop automatically when no longer required.

Scale automatically based on workload processing needs.

Question 6

There are different types of compute. Which ones?

Answer 6

Azure Machine Learning supports multiple types of compute for experimentation and training, and for production inferencing. This enables you to select the most appropriate type of compute target for your particular needs.

Local compute

Training clusters

Inference clusters

Attached compute

Question 7

What is local compute

Answer 7

You can specify a local compute target for most processing tasks in AML. This runs the experiment on the same compute target as the code used to initiate the experiment, which may be your physical workstation or a virtual machine such as an Azure Machine Learning compute instance on which you are running a notebook.

Local compute is generally a great choice during development and testing with low to moderate volumes of data

Question 8

What is training clusters

Answer 8

For training workloads with high scalability requirements, you can use Azure Machine Learning training clusters; which are multi-node clusters of Virtual Machines that automatically scale up or down to meet demand. This is a cost-effective way to run experiments that need to handle large volumes of data or use parallel processing to distribute the workload and reduce the time it takes to run

Question 9

What is the inference clusters

Answer 9

To deploy trained models as production services, you can use Azure Machine Learning inference clusters, which use containerization technologies to enable rapid initialization of compute for on-demand inferencing

Question 10

What is attached compute

Answer 10

If you already use an Azure-based compute environment for data science, such as virtual machine or an Azure Databricks cluster, you can attach it to your AML workspace and use it as a compute target for certain types of workload

Question 11

What is the difference between a managed compute target and an unmanaged compute target

Answer 11

A managed compute target is one that is managed by Azure Machine Learning, such as an Azure Machine Learning training cluster

An unmanaged compute target is one that is defined and managed outside of the Azure Machine Learning workspace; for example, an Azure virtual machine or an Azure Databricks cluster

Question 12

What do you need for running workloads such as experiments?

Answer 12

You need to create an environment and compute targets in your workspace

Question 13

You’re using the Azure Machine Learning Python SDK to run experiments. You need to create an environment from a Conda configuration (.yml) file. Which method of the Environment class should you use?

create

create_from_conda_specification

create_from_existing_conda_environment

Answer 13

create_from_conda_specification
That is correct. Use the create_from_conda_specification method to create an environment from a configuration file. The create method requires you to explicitly specify conda and pip packages, and the create_from_existing_conda_environment requires an existing environment on the computer.

create_from_existing_conda_environment
That is incorrect. create_from_existing_conda_environment requires an existing environment on the computer.

Question 14

You must create a compute target for training experiments that require a graphical processing unit (GPU). You want to be able to scale the compute so that multiple nodes are started automatically as required. Which kind of compute target should you create?

Compute Instance

Training Cluster

Inference Cluster

Answer 14

Training Cluster
That is correct. Use a training cluster to create multiple nodes of GPU-enabled VMs that are started automatically as needed.