Skip to content

Latest commit

 

History

History
535 lines (356 loc) · 19.5 KB

how-to-manage-models.md

File metadata and controls

535 lines (356 loc) · 19.5 KB
Raw
title titleSuffix description services author ms.author ms.reviewer ms.service ms.subservice ms.date ms.topic ms.custom
Register and work with models
Azure Machine Learning
Learn how to register and work with different model types in Azure Machine Learning (such as custom, MLflow, and Triton).
machine-learning
fkriti
kritifaujdar
larryfr
machine-learning
mlops
06/16/2023
conceptual
cli-v2, sdk-v2, devx-track-azurecli, update-code

Work with models in Azure Machine Learning

[!INCLUDE dev v2]

Azure Machine Learning allows you to work with different types of models. In this article, you learn about using Azure Machine Learning to work with different model types, such as custom, MLflow, and Triton. You also learn how to register a model from different locations, and how to use the Azure Machine Learning SDK, the user interface (UI), and the Azure Machine Learning CLI to manage your models.

Tip

If you have model assets created that use the SDK/CLI v1, you can still use those with SDK/CLI v2. Full backward compatibility is provided. All models registered with the V1 SDK are assigned the type custom.

Prerequisites

Additionally, you will need to:

Azure CLI

Python SDK

  • Install the Azure Machine Learning SDK for Python

    pip install azure-ai-ml azure-identity

Supported paths

When you provide a model you want to register, you'll need to specify a path parameter that points to the data or job location. Below is a table that shows the different data locations supported in Azure Machine Learning and examples for the path parameter:

Location Examples
A path on your local computer mlflow-model/model.pkl
A path on an Azure Machine Learning Datastore azureml://datastores/<datastore-name>/paths/<path_on_datastore>
A path from an Azure Machine Learning job azureml://jobs/<job-name>/outputs/<output-name>/paths/<path-to-model-relative-to-the-named-output-location>
A path from an MLflow job runs:/<run-id>/<path-to-model-relative-to-the-root-of-the-artifact-location>
A path from a Model Asset in Azure Machine Learning Workspace azureml:<model-name>:<version>
A path from a Model Asset in Azure Machine Learning Registry azureml://registries/<registry-name>/models/<model-name>/versions/<version>

Supported modes

When you run a job with model inputs/outputs, you can specify the mode - for example, whether you would like the model to be read-only mounted or downloaded to the compute target. The table below shows the possible modes for different type/mode/input/output combinations:

Type Input/Output upload download ro_mount rw_mount direct
custom file Input
custom folder Input ✓ ✓ ✓
mlflow Input ✓ ✓
custom file Output ✓ ✓ ✓
custom folder Output ✓ ✓ ✓
mlflow Output ✓ ✓ ✓

Follow along in Jupyter Notebooks

You can follow along this sample in a Jupyter Notebook. In the azureml-examples repository, open the notebook: model.ipynb.

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.

The code snippets in this section cover how to:

  • Register your model as an asset in Machine Learning by using the CLI.
  • Register your model as an asset in Machine Learning by using the SDK.
  • Register your model as an asset in Machine Learning by using the UI.

These snippets use custom and mlflow.

  • custom is a type that refers to a model file or folder trained with a custom standard not currently supported by Azure Machine Learning.
  • mlflow is a type that refers to a model trained with mlflow. MLflow trained models are in a folder that contains the MLmodel file, the model file, the conda dependencies file, and the requirements.txt file.

Connect to your workspace

First, let's connect to Azure Machine Learning workspace where we are going to work on.

Azure CLI

az account set --subscription <subscription>
az configure --defaults workspace=<workspace> group=<resource-group> location=<location>

Python SDK

The 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. In this section, we'll connect to the workspace in which you'll perform deployment tasks.

  1. Import the required libraries:

    from azure.ai.ml import MLClient, Input
from azure.ai.ml.entities import Model
from azure.ai.ml.constants import AssetTypes
from azure.identity import DefaultAzureCredential

  2. Configure workspace details and get a handle to the workspace:

    subscription_id = "<SUBSCRIPTION_ID>"
resource_group = "<RESOURCE_GROUP>"
workspace = "<AML_WORKSPACE_NAME>"

ml_client = MLClient(DefaultAzureCredential(), subscription_id, resource_group, workspace)

Register your model as an asset in Machine Learning by using the CLI

Use the following tabs to select where your model is located.

Local model

:::code language="yaml" source="~/azureml-examples-main/cli/assets/model/local-file.yml":::

az ml model create -f <file-name>.yml

For a complete example, see the model YAML.

Datastore

You can create a model from a cloud path by using any one of the following supported URI formats.

az ml model create --name my-model --version 1 --path azureml://datastores/myblobstore/paths/models/cifar10/cifar.pt

The examples use the shorthand azureml scheme for pointing to a path on the datastore by using the syntax azureml://datastores/<datastore-name>/paths/<path_on_datastore>.

For a complete example, see the CLI reference.

Job output

You have two options here. You can use the MLflow run URI format, or you can use the azureml job URI format.

MLflow

This option is optimized for MLflow users, who are likely already familiar with the MLflow run URI format. This option helps you create a model from artifacts in the default artifact location (where all MLflow-logged models and artifacts are located). This establishes a lineage between a registered model and the run the model came from.

Format: runs:/<run-id>/<path-to-model-relative-to-the-root-of-the-artifact-location>

Example: runs:/<run-id>/model/

az ml model create --name my-model --version 1 --path runs:/<run-id>/model/ --type mlflow_model

azureml job

This option is an azureml job reference URI format, which helps you register a model from artifacts in any of the job's outputs. This format is aligned with the existing azureml datastore reference URI format, and also supports referencing artifacts from named outputs of the job (not just the default artifact location). You can establish a lineage between a registered model and the job it was trained from, if you didn't directly register your model within the training script by using MLflow.

Format: azureml://jobs/<job-name>/outputs/<output-name>/paths/<path-to-model-relative-to-the-named-output-location>

Examples:

  • Default artifact location: azureml://jobs/<run-id>/outputs/artifacts/paths/model/
    • This is equivalent to runs:/<run-id>/model/.
    • artifacts is the reserved keyword to refer to the output that represents the default artifact location.
  • From a named output directory: azureml://jobs/<run-id>/outputs/trained-model
  • From a specific file or folder path within the named output directory:
    • azureml://jobs/<run-id>/outputs/trained-model/paths/cifar.pt
    • azureml://jobs/<run-id>/outputs/checkpoints/paths/model/

Saving model from a named output:

az ml model create --name my-model --version 1 --path azureml://jobs/<run-id>/outputs/trained-model

For a complete example, see the CLI reference.

Register your model as an asset in Machine Learning by using the SDK

Use the following tabs to select where your model is located.

Local model

[!notebook-python[] (~/azureml-examples-main/sdk/python/assets/model/model.ipynb?name=file_model)]

Datastore

You can create a model from a cloud path by using any one of the following supported URI formats.

[!notebook-python[] (~/azureml-examples-main/sdk/python/assets/model/model.ipynb?name=cloud_model)]

The examples use the shorthand azureml scheme for pointing to a path on the datastore by using the syntax azureml://datastores/${{datastore-name}}/paths/${{path_on_datastore}}.

Job output

You have two options here. You can use the MLflow run URI format, or you can use the azureml job URI format.

MLflow

This option is optimized for MLflow users, who are likely already familiar with the MLflow run URI format. This option helps you create a model from artifacts in the default artifact location (where all MLflow-logged models and artifacts are located). This establishes a lineage between a registered model and the run the model came from.

Format: runs:/<run-id>/<path-to-model-relative-to-the-root-of-the-artifact-location>

Example: runs:/<run-id>/model/

from azure.ai.ml.entities import Model
from azure.ai.ml.constants import ModelType

run_model = Model(
    path="runs:/<run-id>/model/"
    name="run-model-example",
    description="Model created from run.",
    type=ModelType.MLFLOW
)

ml_client.models.create_or_update(run_model)

azureml job

This option is an azureml job reference URI format, which helps you register a model from artifacts in any of the job's outputs. This format is aligned with the existing azureml datastore reference URI format, and also supports referencing artifacts from named outputs of the job (not just the default artifact location). You can establish a lineage between a registered model and the job it was trained from, if you didn't directly register your model within the training script by using MLflow.

Format: azureml://jobs/<job-name>/outputs/<output-name>/paths/<path-to-model-relative-to-the-named-output-location>

Examples:

  • Default artifact location: azureml://jobs/<run-id>/outputs/artifacts/paths/model/
    • This is equivalent to runs:/<run-id>/model/.
    • artifacts is the reserved keyword to refer to the output that represents the default artifact location.
  • From a named output directory: azureml://jobs/<run-id>/outputs/trained-model
  • From a specific file or folder path within the named output directory:
    • azureml://jobs/<run-id>/outputs/trained-model/paths/cifar.pt
    • azureml://jobs/<run-id>/outputs/checkpoints/paths/model/

Saving model from a named output:

[!notebook-python[] (~/azureml-examples-main/sdk/python/assets/model/model.ipynb?name=run_model)]

For a complete example, see the model notebook.

Register your model as an asset in Machine Learning by using the UI

To create a model in Machine Learning, from the UI, open the Models page. Select Register model, and select where your model is located. Fill out the required fields, and then select Register.

:::image type="content" source="./media/how-to-manage-models/register-model-local.png" alt-text="Screenshot of the UI to register a model." lightbox="./media/how-to-manage-models/register-model-local.png":::

Manage models

The SDK and CLI (v2) also allow you to manage the lifecycle of your Azure Machine Learning model assets.

List

List all the models in your workspace:

Azure CLI

az ml model list

Python SDK

models = ml_client.models.list()
for model in models:
    print(model.name)

List all the model versions under a given name:

Azure CLI

az ml model list --name run-model-example

Python SDK

models = ml_client.models.list(name="run-model-example")
for model in models:
    print(model.version)

Show

Get the details of a specific model:

Azure CLI

az ml model show --name run-model-example --version 1

Python SDK

model_example = ml_client.models.get(name="run-model-example", version="1")
print(model_example)

Update

Update mutable properties of a specific model:

Azure CLI

az ml model update --name  run-model-example --version 1 --set description="This is an updated description." --set tags.stage="Prod"

Python SDK

model_example.description="This is an updated description."
model_example.tags={"stage":"Prod"}
ml_client.models.create_or_update(model=model_example)

Important

For model, only description and tags can be updated. All other properties are immutable; if you need to change any of those properties you should create a new version of the model.

Archive

Archiving a model will hide it by default from list queries (az ml model list). You can still continue to reference and use an archived model in your workflows. You can archive either all versions of a model or only a specific version.

If you don't specify a version, all versions of the model under that given name will be archived. If you create a new model version under an archived model container, that new version will automatically be set as archived as well.

Archive all versions of a model:

Azure CLI

az ml model archive --name run-model-example

Python SDK

ml_client.models.archive(name="run-model-example")

Archive a specific model version:

Azure CLI

az ml model archive --name run-model-example --version 1

Python SDK

ml_client.models.archive(name="run-model-example", version="1")

Use model for training

The SDK and CLI (v2) also allow you to use a model in a training job as an input or output.

Use model as input in a job

Azure CLI

Create a job specification YAML file (<file-name>.yml). Specify in the inputs section of the job:

  1. The type; whether the model is a mlflow_model,custom_model or triton_model.
  2. The path of where your data is located; can be any of the paths outlined in the Supported Paths section.

:::code language="yaml" source="~/azureml-examples-main/cli/jobs/basics/hello-model-as-input.yml":::

Next, run in the CLI

az ml job create -f <file-name>.yml

For a complete example, see the model GitHub repo.

Python SDK

The Input class allows you to define:

  1. The type; whether the model is a mlflow_model,custom_model or triton_model.
  2. The path of where your data is located; can be any of the paths outlined in the Supported Paths section.
from azure.ai.ml import command
from azure.ai.ml.entities import Model
from azure.ai.ml import Input
from azure.ai.ml.constants import AssetTypes
from azure.ai.ml import MLClient

# Possible Asset Types for Data:
# AssetTypes.MLFLOW_MODEL
# AssetTypes.CUSTOM_MODEL
# AssetTypes.TRITON_MODEL

# Possible Paths for Model:
# Local path: mlflow-model/model.pkl
# Azure Machine Learning Datastore: azureml://datastores/<datastore-name>/paths/<path_on_datastore>
# MLflow run: runs:/<run-id>/<path-to-model-relative-to-the-root-of-the-artifact-location>
# Job: azureml://jobs/<job-name>/outputs/<output-name>/paths/<path-to-model-relative-to-the-named-output-location>
# Model Asset: azureml:<my_model>:<version>

my_job_inputs = {
    "input_model": Input(type=AssetTypes.MLFLOW_MODEL, path="mlflowmodel")
}

job = command(
    code="./src",  # local path where the code is stored
    command="ls ${{inputs.input_model}}",
    inputs=my_job_inputs,
    environment="AzureML-sklearn-0.24-ubuntu18.04-py37-cpu:9",
    compute="cpu-cluster",
)

# submit the command
returned_job = ml_client.jobs.create_or_update(job)
# get a URL for the status of the job
returned_job.services["Studio"].endpoint

Use model as output in a job

In your job you can write model to your cloud-based storage using outputs.

Azure CLI

Create a job specification YAML file (<file-name>.yml), with the outputs section populated with the type and path of where you would like to write your data to:

:::code language="yaml" source="~/azureml-examples-main/cli/jobs/basics/hello-model-as-output.yml":::

Next create a job using the CLI:

az ml job create --file <file-name>.yml

For a complete example, see the model GitHub repo.

Python SDK

from azure.ai.ml import command
from azure.ai.ml.entities import Model
from azure.ai.ml import Input, Output
from azure.ai.ml.constants import AssetTypes

# Possible Asset Types for Model:
# AssetTypes.MLFLOW_MODEL
# AssetTypes.CUSTOM_MODEL
# AssetTypes.TRITON_MODEL

# Possible Paths for Model:
# Local path: mlflow-model/model.pkl
# Azure Machine Learning Datastore: azureml://datastores/<datastore-name>/paths/<path_on_datastore>
# MLflow run: runs:/<run-id>/<path-to-model-relative-to-the-root-of-the-artifact-location>
# Job: azureml://jobs/<job-name>/outputs/<output-name>/paths/<path-to-model-relative-to-the-named-output-location>
# Model Asset: azureml:<my_model>:<version>

my_job_inputs = {
    "input_model": Input(type=AssetTypes.MLFLOW_MODEL, path="mlflow-model"),
    "input_data": Input(type=AssetTypes.URI_FILE, path="./mlflow-model/input_example.json"),
}

my_job_outputs = {
    "output_folder": Output(type=AssetTypes.CUSTOM_MODEL)
}

job = command(
    code="./src",  # local path where the code is stored
    command="python load_write_model.py --input_model ${{inputs.input_model}} --output_folder ${{outputs.output_folder}}",
    inputs=my_job_inputs,
    outputs=my_job_outputs,
    environment="AzureML-sklearn-0.24-ubuntu18.04-py37-cpu:9",
    compute="cpu-cluster",
)

# submit the command
returned_job = ml_client.create_or_update(job)
# get a URL for the status of the job
returned_job.services["Studio"].endpoint

Next steps