Edit

Share via

How Azure Machine Learning works: resources and assets

APPLIES TO: Azure CLI ml extension v2 (current) Python SDK azure-ai-ml v2 (current)

This article applies to the second version of the Azure Machine Learning CLI and Python SDK v2. For version one (v1), see How Azure Machine Learning works: Architecture and concepts (v1).

Azure Machine Learning includes several resources and assets to enable you to perform your machine learning tasks. These resources and assets are needed to run any job.

  • Resources: setup or infrastructural resources needed to run a machine learning workflow. Resources include:
  • Assets: created with Azure Machine Learning commands or as part of a training/scoring run. Assets are versioned and can be registered in the Azure Machine Learning workspace. They include:

This document provides a quick overview of these resources and assets.

Prerequisites

To use the Python SDK code examples in this article:

  1. Install the Python SDK v2.

  2. Create a connection to your Azure Machine Learning subscription. The examples all rely on ml_client. To create a workspace, the connection doesn't need a workspace name, because you might not have one yet. All other examples in this article require that the workspace name is included in the connection.

    # Import required libraries.
from azure.ai.ml import MLClient
from azure.ai.ml.entities import Workspace
from azure.identity import DefaultAzureCredential
from azure.ai.ml.entities import AmlCompute

# Enter details of your subscription.
subscription_id = "<SUBSCRIPTION_ID>"
resource_group = "<RESOURCE_GROUP>"

# Get a handle to the subscription. (Use this if you haven't created a workspace yet.)
ml_client = MLClient(DefaultAzureCredential(), subscription_id, resource_group)

# All other examples in this article require that the connection include a workspace name.
workspace_name = "<WORKSPACE_NAME>"
ml_client = ml_client = MLClient(DefaultAzureCredential(), subscription_id, resource_group, workspace_name)

Workspace

A workspace is the top-level resource for Azure Machine Learning, providing a centralized place to work with all the artifacts you create when you use Azure Machine Learning. The workspace keeps a history of all jobs, including logs, metrics, output, and a snapshot of your scripts. The workspace stores references to resources like datastores and compute. It also holds all assets, like models, environments, components, and data assets.

Create a workspace

To create a workspace by using Python SDK v2, you can use the following code:

APPLIES TO: Python SDK azure-ai-ml v2 (current)

# Specify the workspace details.
ws = Workspace(
    name="my_workspace",
    location="eastus",
    display_name="My workspace",
    description="This example shows how to create a workspace",
    tags=dict(purpose="demo"),
)
# Use MLClient to connect to the subscription and resource group and create the workspace.
ml_client.workspaces.begin_create(ws)

For more ways to create an Azure Machine Learning workspace by using SDK v2, see this Jupyter notebook.

For more detailed information about creating a workspace, see Manage Azure Machine Learning workspaces in the portal or with the Python SDK (v2).

Compute

A compute is a designated compute resource where you run your job or host your endpoint. Azure Machine Learning supports the following types of compute:

  • Compute instance. A fully configured and managed development environment in the cloud. You can use the instance as a training or inference compute for development and testing. It's similar to a virtual machine in the cloud.
  • Compute cluster. A managed-compute infrastructure that enables you to easily create a cluster of CPU or GPU compute nodes in the cloud.
  • Serverless compute. A compute cluster you access on the fly. When you use serverless compute, you don't need to create your own cluster. All compute lifecycle management is offloaded to Azure Machine Learning.
  • Inference cluster. Used to deploy trained machine learning models to Azure Kubernetes Service (AKS). You can create an Azure Kubernetes Service cluster from your Azure Machine Learning workspace, or attach an existing AKS cluster.
  • Attached compute. You can attach your own compute resources to your workspace and use them for training and inference.

Create a compute resource

To create a compute cluster by using Python SDK v2, you can use the following code:

APPLIES TO: Python SDK azure-ai-ml v2 (current)

cluster_basic = AmlCompute(
    name="basic-example",
    type="amlcompute",
    size="STANDARD_DS3_v2",
    location="westus",
    min_instances=0,
    max_instances=2,
    idle_time_before_scale_down=120,
)
ml_client.begin_create_or_update(cluster_basic)

For more ways to create compute by using SDK v2, see this Jupyter notebook.

For more detailed information about creating compute, see:

Datastore

Azure Machine Learning datastores securely keep the connection information for your data storage on Azure, so you don't have to code it in your scripts. You can register and create a datastore to easily connect to your storage account, and access the data in your underlying storage service. CLI v2 and SDK v2 support the following types of cloud-based storage services:

  • Azure Blob container
  • Azure file share
  • Azure Data Lake Storage
  • Azure Data Lake Storage Gen2

Create a datastore

To create a datastore by using Python SDK v2, you can use the following code:

APPLIES TO: Python SDK azure-ai-ml v2 (current)

import AzureBlobDatastore

blob_datastore1 = AzureBlobDatastore(
    name="blob_example",
    description="Datastore pointing to a blob container.",
    account_name="mytestblobstore",
    container_name="data-container",
    credentials={
        "account_key": "XXXxxxXXXxXXXXxxXXXXXxXXXXXxXxxXxXXXxXXXxXXxxxXXxxXXXxXxXXXxxXxxXXXXxxxxxXXxxxxxxXXXxXXX"
    },
)
ml_client.create_or_update(blob_datastore1)

For more ways to create datastores by using SDK v2, see this Jupyter notebook.

To learn more about using a datastore, see Create and manage data assets.

Model

Azure Machine Learning models consist of one or more binary files that represent a machine learning model and any corresponding metadata. Models can be created from a local or remote file or directory. For remote locations https, wasbs and azureml locations are supported. The created model is tracked in the workspace under the specified name and version. Azure Machine Learning supports three types of storage formats for models:

  • custom_model
  • mlflow_model
  • triton_model

Create a model in the model registry

Model registration allows you to store and version your models in the Azure cloud, in your workspace. The model registry helps you organize and keep track of your trained models.

For more information on how to create models in the registry, see Work with models in Azure Machine Learning.

Environment

An Azure Machine Learning environment is an encapsulation of the environment where your machine learning task happens. It specifies the software packages, environment variables, and software settings for your training and scoring scripts. Environments are managed and versioned entities within your Machine Learning workspace. Environments enable reproducible, auditable, and portable machine learning workflows across various computes.

Types of environments

Azure Machine Learning supports two types of environments: curated and custom.

Curated environments are provided by Azure Machine Learning and are available in your workspace by default. Intended to be used as is, they contain collections of Python packages and settings to help you get started with various machine learning frameworks. These precreated environments also enable faster deployment time. For a full list, see the curated environments article.

In custom environments, you're responsible for setting up your environment and installing packages or any other dependencies that your training or scoring script needs on the compute. Azure Machine Learning allows you to create your own environment by using:

  • A Docker image.
  • A base Docker image with a Conda YAML file for further customizations.
  • A Docker build context.

Create an Azure Machine Learning custom environment

For information about creating an environment by using Python SDK v2, see Create an environment.

For more ways to create custom environments by using SDK v2, see this Jupyter notebook.

For more information about environments, see Create and manage environments in Azure Machine Learning.

Data

Azure Machine Learning allows you to work with different types of data:

  • URIs (a location in local or cloud storage)
    • uri_folder
    • uri_file
  • Tables (a tabular data abstraction)
    • mltable
  • Primitives
    • string
    • boolean
    • number

For most scenarios, you use URIs (uri_folder and uri_file) to a location in storage that you can easily map to the filesystem of a compute node in a job by either mounting or downloading the storage to the node.

mltable is an abstraction for tabular data that's used for AutoML jobs, parallel jobs, and some advanced scenarios. If you're just starting to use Azure Machine Learning and aren't using AutoML, we strongly encourage you to start with URIs.

Component

An Azure Machine Learning component is a self-contained piece of code that completes one step in a machine learning pipeline. Components are the building blocks of advanced machine learning pipelines. Components can do tasks like data processing, model training, and model scoring. A component is analogous to a function: it has a name and parameters, expects input, and returns output.

Related content