models parameter is not documented/add more info on models parameter

[lordlinus]

[Enter feedback here]

models take an array of models and links/example on how to create this model using MLFlow or other means
e.g. below is an example of how to create an azure model to be used in Model.deploy

model_image, azure_model = mlflow.azureml.build_image(model_uri=model_uri,
workspace=workspace,
model_name=model_name,
image_name=image_name,
description="some example description",
synchronous=True)

service = Model.deploy(workspace=workspace,
name=prod_webservice_name,
models=[azure_model],
inference_config=inference_config,
deployment_config=prod_webservice_deployment_config,
deployment_target = aks_target,
overwrite=True)

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: f3e0f3c7-6241-c84e-e655-03e5c3ed4780
• Version Independent ID: 6e6bb0d2-2ab8-651a-190d-90c198eca17c
• Content: azureml.core.model.InferenceConfig class - Azure Machine Learning Python
• Content Source: AzureML-Docset/stable/docs-ref-autogen/azureml-core/azureml.core.model.InferenceConfig.yml
• Service: machine-learning
• Sub-service: core
• GitHub Login: @DebFro
• Microsoft Alias: debfro

[dem108]

Hi @lordlinus, sorry for taking long to come back to your question. We're designing the improved experience to support multi-model or ensemble model. Until then, the workaround is to include multiple models under a directory, and register the directory as a parent model. In your scoring script, use the environment variable AZUREML_MODEL_DIR to get the path to the model root folder. You can refer to child directories, and use them for model ensemble. For more information, please see use more than one model.