Azure Machine Learning (AML) is a cloud-based platform from Microsoft that is used to train, deploy, automate, and manage machine learning models. A key feature of AML is the ability to trigger jobs, such as training and running experiments and pipeline, directly from your development workflow. This can be done using AML’s integrations with Azure DevOps and GitHub.
Azure Machine Learning Job
An AML job is an action you want to run on the AML platform. This could be data pre-processing, model training, scoring, or validation among others. These jobs can be executed locally, on Azure Machine Learning Compute, or on other supported compute targets.
Jobs can be triggered manually, scheduled, or even automated to execute when certain conditions are met.
Triggering an AML Job from Azure DevOps
Azure DevOps is a Microsoft product that offers a full DevOps tool chain for developing and deploying software. Azure DevOps includes Azure Pipelines, which are used to automatically build, test, and deploy applications.
You can trigger an AML job by creating a pipeline in Azure DevOps and setting it up to call the AML training script.
Here’s a basic example of how a pipeline could be set up to trigger an AML job:
trigger:
– master
pool:
vmImage: ‘ubuntu-latest’
steps:
– task: UsePythonVersion@0
inputs:
versionSpec: ‘3.6’
addToPath: true
– script: pip install azureml-sdk
displayName: ‘Install dependencies’
– script: python train.py
displayName: ‘Train model’
In this example, the AML job is triggered any time a change is committed to the ‘master’ branch. The job uses Python 3.6 and installs the Azure Machine Learning SDK. The training script is executed with the ‘python train.py’ command.
Triggering an AML Job from GitHub
GitHub is a web-based hosting service for version controlling using Git. It’s used by many developers to collaborate on projects.
You can use GitHub Actions, a new feature of GitHub, to trigger an AML job. GitHub Actions are a flexible way to automate your workflow.
Here is an example of setting up a GitHub Action to trigger an AML job:
name: Train
on: [push]
jobs:
build:
runs-on: ubuntu-latest
steps:
– uses: actions/checkout@v2
– name: Set up Python 3.6
uses: actions/setup-python@v1
with:
python-version: 3.6
– name: Install dependencies
run: pip install azureml-sdk
– name: Train model
run: python train.py
In this example, the AML job is triggered any time code is pushed to the repository. The job sets up Python 3.6, installs the Azure Machine Learning SDK, and then runs a Python script to train the model.
In summary, it is easy to configure both Azure DevOps and GitHub to trigger Azure Machine Learning jobs. This allows you to integrate machine learning into your existing workflows, and makes it easier to automate your machine learning tasks. All you need to do is write your training scripts, configure your triggers, and you’re good to go!
Practice Test
True or False: Azure Machine Learning can be integrated with Azure DevOps.
- True
- False
Answer: True
Explanation: Azure Machine Learning can be integrated with Azure DevOps to automate the building, training, and deployment of machine learning models.
In Azure, the primary unit of work for machine learning training is a ___?
- A. Job
- B. Project
- C. Resource Group
- D. Container
Answer: A. Job
Explanation: In Azure Machine Learning, the primary unit of work for training is a job which can be triggered to start.
True or False: Jobs in Azure Machine Learning cannot be triggered from GitHub.
- True
- False
Answer: False
Explanation: Jobs can be triggered from different platforms including GitHub by integrating the platform with Azure Machine Learning through APIs.
What is the main use of Azure DevOps in the context of Azure Machine Learning?
- A. Create and manage databases
- B. Troubleshoot network problems
- C. Automate the building, training and deployment of machine learning models
- D. All of the above
Answer: C. Automate the building, training and deployment of machine learning models
Explanation: Azure DevOps is mainly used to automate the entire lifecycle of machine learning models including building, training and deployment of models in Azure Machine Learning.
True or False: Triggers in Azure Machine Learning are automatic and cannot be manually controlled.
- True
- False
Answer: False
Explanation: Triggers in Azure Machine Learning can be manually controlled to start and stop whenever needed.
Which component of Azure DevOps is used to automate the deployment of Azure Machine Learning models?
- A. Azure Boards
- B. Azure Repos
- C. Azure Pipelines
- D. Azure Test Plans
Answer: C. Azure Pipelines
Explanation: Azure Pipelines, a component of Azure DevOps, provides build and release services to support continuous integration and delivery of your apps.
True or False: Azure Machine Learning jobs require manual intervention at every step.
- True
- False
Answer: False
Explanation: Using Azure DevOps, the building, training and deployment of Azure Machine Learning models can be fully automated, requiring minimal manual intervention.
How can you trigger an Azure Machine Learning job from GitHub?
- A. Using Azure APIs
- B. Using Python SDK
- C. Not possible to trigger from GitHub
- D. Both A and B
Answer: D. Both A and B
Explanation: Azure provides APIs and Python SDK which can be used to integrate with GitHub to trigger jobs in Azure Machine Learning.
True or False: It is not possible to connect Azure DevOps with GitHub.
- True
- False
Answer: False
Explanation: It is possible to connect Azure DevOps with GitHub for continuous integration and continuous deployment.
What is the role of a trigger in Azure Machine Learning jobs?
- A. To initiate a job
- B. To stop a job
- C. To debug a job
- D. All of the above
Answer: A. To initiate a job
Explanation: A trigger in Azure Machine Learning is used to start or initiate a job.
Interview Questions
What is an Azure Machine Learning job?
An Azure Machine Learning job is a unit of compute in Azure Machine Learning that performs a specific task. It can be used for training, scoring, or evaluating models.
How can you trigger an Azure Machine Learning job from Azure DevOps?
You can trigger an Azure Machine Learning job from Azure DevOps by using Azure Pipelines. This involves setting up a pipeline that includes a step to run the job, and then triggering the pipeline either manually or with an event, such as a code commit.
How to trigger an Azure Machine Learning job from GitHub?
You can trigger an Azure Machine Learning job from GitHub using GitHub Actions. This involves setting up an action that includes the Azure login and Azure Machine Learning run steps and then defining a trigger for the action, such as a push to a specific branch.
What is Azure DevOps?
Azure DevOps is a platform from Microsoft that provides various services for software development lifecycle such as Azure Boards, Azure Repos, Azure Pipelines, Azure Test Plans, and Azure Artifacts.
What is GitHub Actions?
GitHub Actions is a feature of GitHub that enables you to create custom software development lifecycle workflows directly in your GitHub repository.
How to monitor an Azure Machine Learning job?
You can monitor an Azure Machine Learning job using the Azure Machine Learning studio or SDK. You can check the job’s progress, view its results, and troubleshoot if necessary.
Can you automate the triggering of an Azure Machine Learning job?
Yes, you can automate the triggering of an Azure Machine Learning job. This often involves using a pipeline in Azure DevOps or an action in GitHub, which can be triggered by events like a code commit or a scheduled time.
Are Azure Pipelines and GitHub Actions free to use?
Both Azure Pipelines and GitHub Actions offer free tiers, but there may be costs associated with higher usage levels or for certain features.
Can you use both Azure DevOps and GitHub together?
Yes, Azure DevOps and GitHub can be used together. For example, you could use Azure Boards from Azure DevOps for tracking work and Azure Pipelines for running CI/CD workflows, while storing your code in a GitHub repository.
What are the main components of Azure Pipelines?
The main components of Azure Pipelines are pipelines, tasks, and jobs. A pipeline defines the complete set of tasks that run in sequence, a task is the smallest unit of work in a pipeline, and a job is a collection of tasks that run on an agent or on the server.
What would be a use-case for triggering an Azure Machine Learning job from Azure DevOps or GitHub?
A use-case for triggering an Azure Machine Learning job from Azure DevOps or GitHub could be part of an automated machine learning workflow. This might involve retraining a model whenever new data is added to the dataset or on a scheduled basis.
What’s the role of an agent in Azure Pipelines?
An agent in Azure Pipelines is a computer that runs a job. Each job is run on one agent. An agent can be hosted by Microsoft or you can manage your own.
How can you debug a failed Azure Machine Learning job?
You can debug a failed Azure Machine Learning job by looking at the job’s logs and metrics in the Azure Machine Learning studio. Additionally, for failures during training, you can use the debugging and analysis features provided by the Azure Machine Learning SDK.
What are the steps to trigger an Azure Machine Learning job from Azure DevOps?
To trigger an Azure Machine Learning job from Azure DevOps, follow these steps:
1. Create an Azure DevOps project or use an existing one.
2. In your project, create a new pipeline.
3. In the pipeline, add a step that runs the Azure Machine Learning job.
4. Define a trigger for the pipeline, such as a code commit or a scheduled time.
5. Save and run the pipeline.
If you were using GitHub actions to trigger Azure Machine Learning jobs, what types of events could be used to start the run?
If you’re using GitHub Actions, several types of events could trigger a run. This includes push events, pull requests, releases, manual workflows, or scheduled times.
extutorials.com