Merge pull request #94064 from amymsft/patch-2

Add compute to tutorial

Author: jborsecnik

articles/machine-learning/how-to-use-batch-endpoint-sdk-v2.md

@@ -50,6 +50,7 @@ The [workspace](concept-workspace.md) is the top-level resource for Azure Machin
     # import required libraries
     from azure.ai.ml import MLClient, Input
     from azure.ai.ml.entities import (
+        AmlCompute,
         BatchEndpoint,
         BatchDeployment,
         Model,
@@ -118,6 +119,16 @@ To create an online endpoint, we'll use `BatchEndpoint`. This class allows user
     ml_client.begin_create_or_update(endpoint)
     ```
 
+## Create batch compute
+
+Batch endpoint runs only on cloud computing resources, not locally. The cloud computing resource is a reusable virtual computer cluster. Run the following code to create an Azure Machine Learning compute cluster. The following examples in this article use the compute created here named `cpu-cluster`.
+
+```python
+compute_name = "cpu-cluster"
+compute_cluster = AmlCompute(name=compute_name, description="amlcompute", min_instances=0, max_instances=5)
+ml_client.begin_create_or_update(compute_cluster)
+```
+
 ## Create a deployment
 
 A deployment is a set of resources required for hosting the model that does the actual inferencing. We'll create a deployment for our endpoint using the `BatchDeployment` class. This class allows user to configure the following key aspects.
@@ -157,7 +168,7 @@ A deployment is a set of resources required for hosting the model that does the
         code_path="./mnist/code/",
         scoring_script="digit_identification.py",
         environment=env,
-        compute="cpu-cluster",
+        compute=compute_name,
         instance_count=2,
         max_concurrency_per_instance=2,
         mini_batch_size=10,
@@ -222,4 +233,4 @@ ml_client.batch_endpoints.begin_delete(name=batch_endpoint_name)
 
 ## Next steps
 
-If you encounter problems using batch endpoints, see [Troubleshooting batch endpoints](how-to-troubleshoot-batch-endpoints.md).
+If you encounter problems using batch endpoints, see [Troubleshooting batch endpoints](how-to-troubleshoot-batch-endpoints.md).