Merge pull request #215416 from santiagxf/santiagxf/aml-batch-defaults

Update how-to-mlflow-batch.md

Author: prmerger-automator[bot]

articles/machine-learning/batch-inference/how-to-deploy-model-custom-output.md

@@ -179,7 +179,8 @@ Follow the next steps to create a deployment using the previous scoring script:
    Then, create the deployment with the following command:
    
    ```azurecli
-   az ml batch-endpoint create -f endpoint.yml
+   DEPLOYMENT_NAME="classifier-xgboost-parquet"
+   az ml batch-deployment create -f endpoint.yml
    ```
 
    # [Azure ML SDK for Python](#tab/sdk)
@@ -205,6 +206,11 @@ Follow the next steps to create a deployment using the previous scoring script:
        retry_settings=BatchRetrySettings(max_retries=3, timeout=300),
        logging_level="info",
    )
+   ```
+   
+   Then, create the deployment with the following command:
+   
+   ```python
    ml_client.batch_deployments.begin_create_or_update(deployment)
    ```
    ---
@@ -259,7 +265,7 @@ For testing our endpoint, we are going to use a sample of unlabeled data located
    # [Azure ML CLI](#tab/cli)
 
    ```azurecli
-   JOB_NAME = $(az ml batch-endpoint invoke --name $ENDPOINT_NAME --input azureml:heart-dataset-unlabeled@latest | jq -r '.name')
+   JOB_NAME = $(az ml batch-endpoint invoke --name $ENDPOINT_NAME --deployment-name $DEPLOYMENT_NAME --input azureml:heart-dataset-unlabeled@latest | jq -r '.name')
    ```
 
    > [!NOTE]

articles/machine-learning/batch-inference/how-to-image-processing-batch.md

@@ -204,7 +204,8 @@ One the scoring script is created, it's time to create a batch deployment for it
    Then, create the deployment with the following command:
    
    ```azurecli
-   az ml batch-endpoint create -f endpoint.yml
+   DEPLOYMENT_NAME="imagenet-classifier-resnetv2"
+   az ml batch-deployment create -f deployment.yml
    ```
 
    # [Azure ML SDK for Python](#tab/sdk)
@@ -232,8 +233,29 @@ One the scoring script is created, it's time to create a batch deployment for it
        logging_level="info",
    )
    ```
+   
+   Then, create the deployment with the following command:
+   
+   ```python
+   ml_client.batch_deployments.begin_create_or_update(deployment)
+   ```
+
+1. Although you can invoke a specific deployment inside of an endpoint, you will usually want to invoke the endpoint itself and let the endpoint decide which deployment to use. Such deployment is named the "default" deployment. This gives you the possibility of changing the default deployment and hence changing the model serving the deployment without changing the contract with the user invoking the endpoint. Use the following instruction to update the default deployment:
+
+   # [Azure ML CLI](#tab/cli)
+   
+   ```bash
+   az ml batch-endpoint update --name $ENDPOINT_NAME --set defaults.deployment_name=$DEPLOYMENT_NAME
+   ```
+   
+   # [Azure ML SDK for Python](#tab/sdk)
+   
+   ```python
+   endpoint.defaults.deployment_name = deployment.name
+   ml_client.batch_endpoints.begin_create_or_update(endpoint)
+   ```
 
-1. At this point, our batch endpoint is ready to be used. 
+1. At this point, our batch endpoint is ready to be used.  
 
 ## Testing out the deployment
 

articles/machine-learning/batch-inference/how-to-mlflow-batch.md

@@ -196,6 +196,7 @@ Follow these steps to deploy an MLflow model to a batch endpoint for running bat
    Then, create the deployment with the following command:
    
    ```bash
+   DEPLOYMENT_NAME="classifier-xgboost-mlflow"
    az ml batch-endpoint create -f endpoint.yml
    ```
 
@@ -225,7 +226,22 @@ Follow these steps to deploy an MLflow model to a batch endpoint for running bat
    > [!NOTE]
    > `scoring_script` and `environment` auto generation only supports `pyfunc` model flavor. To use a different flavor, see [Using MLflow models with a scoring script](#using-mlflow-models-with-a-scoring-script).
 
-6. At this point, our batch endpoint is ready to be used. 
+6. Although you can invoke a specific deployment inside of an endpoint, you will usually want to invoke the endpoint itself and let the endpoint decide which deployment to use. Such deployment is named the "default" deployment. This gives you the possibility of changing the default deployment and hence changing the model serving the deployment without changing the contract with the user invoking the endpoint. Use the following instruction to update the default deployment:
+
+   # [Azure ML CLI](#tab/cli)
+   
+   ```bash
+   az ml batch-endpoint update --name $ENDPOINT_NAME --set defaults.deployment_name=$DEPLOYMENT_NAME
+   ```
+   
+   # [Azure ML SDK for Python](#tab/sdk)
+   
+   ```python
+   endpoint.defaults.deployment_name = deployment.name
+   ml_client.batch_endpoints.begin_create_or_update(endpoint)
+   ```
+
+7. At this point, our batch endpoint is ready to be used. 
 
 ## Testing out the deployment
 

articles/machine-learning/batch-inference/how-to-nlp-processing-batch.md

@@ -181,7 +181,8 @@ One the scoring script is created, it's time to create a batch deployment for it
    Then, create the deployment with the following command:
    
    ```bash
-   az ml batch-endpoint create -f endpoint.yml
+   DEPLOYMENT_NAME="text-summarization-hfbart"
+   az ml batch-deployment create -f endpoint.yml
    ```
 
    # [Azure ML SDK for Python](#tab/sdk)
@@ -208,15 +209,35 @@ One the scoring script is created, it's time to create a batch deployment for it
        retry_settings=BatchRetrySettings(max_retries=3, timeout=3000),
        logging_level="info",
    )
+   ```
+   
+   Then, create the deployment with the following command:
 
+   ```python
    ml_client.batch_deployments.begin_create_or_update(deployment)
    ```
    ---
 
    > [!IMPORTANT]
    > You will notice in this deployment a high value in `timeout` in the parameter `retry_settings`. The reason for it is due to the nature of the model we are running. This is a very expensive model and inference on a single row may take up to 60 seconds. The `timeout` parameters controls how much time the Batch Deployment should wait for the scoring script to finish processing each mini-batch. Since our model runs predictions row by row, processing a long file may take time. Also notice that the number of files per batch is set to 1 (`mini_batch_size=1`). This is again related to the nature of the work we are doing. Processing one file at a time per batch is expensive enough to justify it. You will notice this being a pattern in NLP processing.
 
-3. At this point, our batch endpoint is ready to be used. 
+3. Although you can invoke a specific deployment inside of an endpoint, you will usually want to invoke the endpoint itself and let the endpoint decide which deployment to use. Such deployment is named the "default" deployment. This gives you the possibility of changing the default deployment and hence changing the model serving the deployment without changing the contract with the user invoking the endpoint. Use the following instruction to update the default deployment:
+
+   # [Azure ML CLI](#tab/cli)
+   
+   ```bash
+   az ml batch-endpoint update --name $ENDPOINT_NAME --set defaults.deployment_name=$DEPLOYMENT_NAME
+   ```
+   
+   # [Azure ML SDK for Python](#tab/sdk)
+   
+   ```python
+   endpoint.defaults.deployment_name = deployment.name
+   ml_client.batch_endpoints.begin_create_or_update(endpoint)
+   ```
+
+4. At this point, our batch endpoint is ready to be used. 
+
 
 ## Considerations when deploying models that process text