This template can be used for easily setting up a data science or machine learning project with automated training and deployment using GitHub Actions and Azure Machine Learning. For a more comprehensive version of this automated pipeline, see the aml-template repository.
The following prerequisites are required to make this repository work:
- Azure subscription
- Contributor access to the Azure subscription
- Owner permissions for the workspace you are trying to automate
- Access to GitHub Actions
If you don’t have an Azure subscription, create a free account before you begin. Try the free or paid version of Azure Machine Learning today.
Create a new repo based off this template, by clicking on the green 'Use this template' button:
A service principal (SP) needs to be generated for authentication and accessing your Azure subscription. We suggest adding a service principal with contributor rights to a new resource group, or to the one where you have deployed your existing Azure Machine Learning workspace.
Navigate to the Azure Portal to find the details of your resource group or workspace. Create a Machine Learning workspace if you haven't already.
Note: If you can't add a role assignment, you are not permissioned with the ability to create a service principal. Contact your IT admin to remedy this. (workspace > 'Access Control(IAM)' > 'Add a role assignment' > 'Add').
Service Principal (SP) Authentication using AZ CLI
Additional SP setup methods can be found here, in the section titled 'Service Principal Authentication'.
You will need the following to create your SP in the AZ CLI:
- {service-principal-name} = pick a new name for your service principal (e.g.workspacename_SPauth)
- {subscription-id} = Workspace > Overview > subscription ID
- {resource-group}= Workspace > Overview > resource group
Install the Azure CLI(recommended) on your computer or use the Cloud CLI
Execute the following command to generate the required credentials for your GitHub Action:
az ad sp create-for-rbac --name {service-principal-name} \
--role contributor \
--scopes /subscriptions/{subscription-id}/resourceGroups/{resource-group} \
--sdk-auth
This will generate the following JSON output:
{
"clientId": "<GUID>",
"clientSecret": "<GUID>",
"subscriptionId": "<GUID>",
"tenantId": "<GUID>",
(...)
}
Add this JSON output as a secret in your GitHub repository with the name AZURE_CREDENTIALS
. (Settings >'Secrets' > add AZURE_CREDENTIALS
).
Note: your secret must be named AZURE_CREDENTIALS
for the GitHub Action to work correctly
More information on GitHub secrets can be found here.
You will need to modify your workspace.json
file
Location: /.cloud/.azure/workspace.json"
file in your repository
Your GitHub Actions will use this to connect to your desired AzureML workspace. You will need the following:
- {name} = same value for your workspace name (as above), or pick a new name for your new workspace. You can also delete the
name
parameter, if you want the action to use the default value, which is the repository name. - {resource_group} = same resource group as above
- {create_workspace}= this stays unchanged
Make your edits to workspace.json
:
{
"name": "your-workspace-name",
"resource_group": "your-resource-group",
"create_workspace": true
}
Once you save your changes to the file, the predefined GitHub workflow that trains and deploys a model on Azure Machine Learning gets triggered.
Check the actions tab to view if your actions have successfully run.
Now you can start modifying the code in the code folder
, so that your model and not the provided sample model gets trained on Azure.
Where required, modify the environment yaml so that the training and deployment environments will have the correct packages installed in the conda environment for your training and deployment.
Upon pushing the changes, actions will kick off your training and deployment run.
Check the actions tab to view if your actions have successfully run.
OPTIONAL: Comment lines 39 to 55 in your "/.github/workflows/train_deploy.yml"
file if you only want to train the model. Uncomment line 7 to 8, if you only want to kick off the workflow when pushing changes to the "/code/"
file.
The log outputs of your action will provide URLs for you to view the resources that have been created in AML. Alternatively, you can visit the Machine Learning Studio to view the progress of your runs, etc.
File/folder | Description |
---|---|
code |
Sample data science source code that will be submitted to Azure Machine Learning to train and deploy machine learning models. |
code/train |
Sample code that is required for training a model on Azure Machine Learning. |
code/train/train.py |
Training script that gets executed on a cluster on Azure Machine Learning. |
code/train/environment.yml |
Conda environment specification, which describes the dependencies of train.py . These packages will be installed inside a Docker image on the Azure Machine Learning compute cluster, when executing your train.py . |
code/train/run_config.yml |
YAML files, which describes the execution of your training run on Azure Machine Learning. This file also references your environment.yml . Please look at the comments in the file for more details. |
code/deploy |
Sample code that is required for deploying a model on Azure Machine Learning. |
code/deploy/score.py |
Inference script that is used to build a Docker image and that gets executed within the container when you send data to the deployed model on Azure Machine Learning. |
code/deploy/environment.yml |
Conda environment specification, which describes the dependencies of score.py . These packages will be installed inside the Docker image that will be used for deploying your model. |
code/test/test.py |
Test script that can be used for testing your deployed webservice. Add a deploy.json to the .cloud/.azure folder and add the following code { "test_enabled": true } to enable tests of your webservice. Change the code according to the tests that zou would like to execute. |
.cloud/.azure |
Configuration files for the Azure Machine Learning GitHub Actions. Please visit the repositories of the respective actions and read the documentation for more details. |
.github/workflows |
Folder for GitHub workflows. The train_deploy.yml sample workflow shows you how your can use the Azure Machine Learning GitHub Actions to automate the machine learning process. |
docs |
Resources for this README. |
CODE_OF_CONDUCT.md |
Microsoft Open Source Code of Conduct. |
LICENSE |
The license for the sample. |
README.md |
This README file. |
SECURITY.md |
Microsoft Security README. |
The template uses the open source Azure certified Actions listed below. Click on the links and read the README files for more details.
- aml-workspace - Connects to or creates a new workspace
- aml-compute - Connects to or creates a new compute target in Azure Machine Learning
- aml-run - Submits a ScriptRun, an Estimator or a Pipeline to Azure Machine Learning
- aml-registermodel - Registers a model to Azure Machine Learning
- aml-deploy - Deploys a model and creates an endpoint for the model
Error message:
Message: ***'error': ***'code': 'MissingSubscriptionRegistration', 'message': "The subscription is not registered to use namespace 'Microsoft.KeyVault'. See https://aka.ms/rps-not-found for how to register subscriptions.", 'details': [***'code': 'MissingSubscriptionRegistration', 'target': 'Microsoft.KeyVault', 'message': "The subscription is not registered to use namespace 'Microsoft.KeyVault'. See https://aka.ms/rps-not-found for how to register subscriptions
Solution:
This error message appears, in case the Azure/aml-workspace
action tries to create a new Azure Machine Learning workspace in your resource group and you have never deployed a Key Vault in the subscription before. We recommend to create an Azure Machine Learning workspace manually in the Azure Portal. Follow the steps on this website to create a new workspace with the desired name. After you have successfully completed the steps, you have to make sure, that your Service Principal has access to the resource group and that the details in your /.cloud/.azure/workspace.json"
file are correct and point to the right workspace and resource group.
MLOps empowers data scientists and machine learning engineers to bring together their knowledge and skills to simplify the process of going from model development to release/deployment. ML Ops enables you to track, version, test, certify and reuse assets in every part of the machine learning lifecycle and provides orchestration services to streamline managing this lifecycle. This allows practitioners to automate the end to end machine Learning lifecycle to frequently update models, test new models, and continuously roll out new ML models alongside your other applications and services.
This repository enables Data Scientists to focus on the training and deployment code of their machine learning project (code
folder of this repository). Once new code is checked into the code
folder of the master branch of this repository the GitHub workflow is triggered and open source Azure Machine Learning actions are used to automatically manage the training through to deployment phases.
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.
When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.