note/include and grammar

Author: Larry

articles/machine-learning/concept-plan-manage-cost.md

@@ -71,10 +71,12 @@ Because these compute pools are inside of Azure's IaaS infrastructure, you can d
 Autoscaling clusters based on the requirements of your workload helps reduce your costs so you only use what you need. 
 AmlCompute clusters are designed to autoscale dynamically based on the requirements of your workload. The cluster can be scaled up to the maximum number of nodes provisioned and within the quota designated for the subscription. As each run completes, the cluster will release nodes and autoscale to your designated minimum node count.
 
+[!INCLUDE [min-nodes-note](../../includes/machine-learning-min-nodes.md)]
+
 In addition to setting the minimum and maximum number of nodes, tweak the amount of time the node is idle before scale down. By default, idle time before scale down is set to 120 seconds.
 
 + If you perform less iterative experimentation, reduce this time to save costs. 
-+ If you perform highly iterative dev/test experimentation, you might need to increase this so you aren't paying for constant scaling up and down after each change to your training script or environment.
++ If you perform highly iterative dev/test experimentation, you might need to increase the time so you aren't paying for constant scaling up and down after each change to your training script or environment.
 
 AmlCompute clusters can be configured for your changing workload requirements in Azure portal, using the [AmlCompute SDK class](https://docs.microsoft.com/python/api/azureml-core/azureml.core.compute.amlcompute.amlcompute?view=azure-ml-py), [AmlCompute CLI](https://docs.microsoft.com/cli/azure/ext/azure-cli-ml/ml/computetarget/create?view=azure-cli-latest#ext-azure-cli-ml-az-ml-computetarget-create-amlcompute), with the [REST APIs](https://github.com/Azure/azure-rest-api-specs/tree/master/specification/machinelearningservices/resource-manager/Microsoft.MachineLearningServices/stable).
 
@@ -86,17 +88,17 @@ az ml computetarget create amlcompute --name testcluster --vm-size Standard_NC6
 
 Much like other Azure compute resources, AmlCompute comes with an inherent [quota (or limit) configuration](how-to-manage-quotas.md#azure-machine-learning-compute). This quota is by VM family (for example, Dv2 series, NCv3 series) and varies by region for each subscription. Subscriptions start with small defaults to get you going, but use this setting to control the amount of Amlcompute resources available to be spun up in your subscription. 
 
-Also configure [workspace level quota by VM family](how-to-manage-quotas.md#workspace-level-quota), for each workspace within a subscription. This allows you to have more granular control on the costs that each workspace might potentially incur and restrict certain VM families. 
+Also configure [workspace level quota by VM family](how-to-manage-quotas.md#workspace-level-quota), for each workspace within a subscription. Doing so allows you to have more granular control on the costs that each workspace might potentially incur and restrict certain VM families. 
 
-To set quotas at the workspace level, start in the [Azure portal](https://portal.azure.com).  Select any workspace in your subscription, and select **Usages + quotas** in the left pane. Then select the **Configure quotas** tab to view the quotas. You need privileges at the subscription scope to set this quota, since it's a setting that affects multiple workspaces.
+To set quotas at the workspace level, start in the [Azure portal](https://portal.azure.com).  Select any workspace in your subscription, and select **Usages + quotas** in the left pane. Then select the **Configure quotas** tab to view the quotas. You need privileges at the subscription scope to set the quota, since it's a setting that affects multiple workspaces.
 
-## Set run auto-termination policies 
+## Set run autotermination policies 
 
-Configure your training runs to limit their duration or to terminate them early in case of certain conditions especially when you are using Azure Machine Learning's built-in Hyperparameter Tuning or Automated Machine Learning capabilities. 
+In some cases, you should configure your training runs to limit their duration or terminate them early. For example, when you are using Azure Machine Learning's built-in hyperparameter tuning or automated machine learning.
 
 Here are a few options that you have:
 * Define a parameter called `max_run_duration_seconds` in your RunConfiguration to control the maximum duration a run can extend to on the compute you choose (either local or remote cloud compute).
-* For [hyperparameter tuning](how-to-tune-hyperparameters.md#early-termination), define an early termination policy from a Bandit policy, a Median stopping policy or a Truncation selection policy. In addition, also use parameters such as `max_total_runs` or `max_duration_minutes` to further control the various hyperparameter sweeps.
+* For [hyperparameter tuning](how-to-tune-hyperparameters.md#early-termination), define an early termination policy from a Bandit policy, a Median stopping policy, or a Truncation selection policy. In addition, also use parameters such as `max_total_runs` or `max_duration_minutes` to further control the various hyperparameter sweeps.
 * For [automated machine learning](how-to-configure-auto-train.md#exit), set similar termination policies using the  `enable_early_stopping` flag. Also use properties such as `iteration_timeout_minutes` and `experiment_timeout_minutes` to control the maximum duration of a run or for the entire experiment.
 
 ## Use low-priority VMs

articles/machine-learning/how-to-set-up-training-targets.md

@@ -15,9 +15,9 @@ ms.custom: seodec18
 # Set up and use compute targets for model training 
 [!INCLUDE [applies-to-skus](../../includes/aml-applies-to-basic-enterprise-sku.md)]
 
-With Azure Machine Learning, you can train your model on a variety of resources or environments, collectively referred to as [__compute targets__](concept-azure-machine-learning-architecture.md#compute-targets). A compute target can be a local machine or a cloud resource, such as an Azure Machine Learning Compute, Azure HDInsight or a remote virtual machine.  You can also create compute targets for model deployment as described in ["Where and how to deploy your models"](how-to-deploy-and-where.md).
+With Azure Machine Learning, you can train your model on a variety of resources or environments, collectively referred to as [__compute targets__](concept-azure-machine-learning-architecture.md#compute-targets). A compute target can be a local machine or a cloud resource, such as an Azure Machine Learning Compute, Azure HDInsight, or a remote virtual machine.  You can also create compute targets for model deployment as described in ["Where and how to deploy your models"](how-to-deploy-and-where.md).
 
-You can create and manage a compute target using the Azure Machine Learning SDK, Azure Machine Learning studio, Azure CLI or Azure Machine Learning VS Code extension. If you have compute targets that were created through another service (for example, an HDInsight cluster), you can use them by attaching them to your Azure Machine Learning workspace.
+You can create and manage a compute target using the Azure Machine Learning SDK, Azure Machine Learning studio, Azure CLI, or Azure Machine Learning VS Code extension. If you have compute targets that were created through another service (for example, an HDInsight cluster), you can use them by attaching them to your Azure Machine Learning workspace.
 
 In this article, you learn how to use various compute targets for model training.  The steps for all compute targets follow the same workflow:
 1. __Create__ a compute target if you don't already have one.
@@ -30,7 +30,7 @@ In this article, you learn how to use various compute targets for model training
 
 ## Compute targets for training
 
-Azure Machine Learning has varying support across different compute targets. A typical model development lifecycle starts with dev/experimentation on a small amount of data. At this stage, we recommend using a local environment. For example, your local computer or a cloud-based VM. As you scale up your training on larger data sets, or do distributed training, we recommend using Azure Machine Learning Compute to create a single- or multi-node cluster that autoscales each time you submit a run. You can also attach your own compute resource, although support for various scenarios may vary as detailed below:
+Azure Machine Learning has varying support across different compute targets. A typical model development lifecycle starts with dev/experimentation on a small amount of data. At this stage, we recommend using a local environment. For example, your local computer or a cloud-based VM. As you scale up your training on larger data sets, or perform distributed training, we recommend using Azure Machine Learning Compute to create a single- or multi-node cluster that autoscales each time you submit a run. You can also attach your own compute resource, although support for various scenarios may vary as detailed below:
 
 [!INCLUDE [aml-compute-target-train](../../includes/aml-compute-target-train.md)]
 
@@ -58,7 +58,7 @@ For more information, see [Train ML Models with estimators](how-to-train-ml-mode
 
 With ML pipelines, you can optimize your workflow with simplicity, speed, portability, and reuse. When building pipelines with Azure Machine Learning, you can focus on your expertise, machine learning, rather than on infrastructure and automation.
 
-ML pipelines are constructed from multiple **steps**, which are distinct computational units in the pipeline. Each step can run independently and use isolated compute resources. This allows multiple data scientists to work on the same pipeline at the same time without over-taxing compute resources, and also makes it easy to use different compute types/sizes for each step.
+ML pipelines are constructed from multiple **steps**, which are distinct computational units in the pipeline. Each step can run independently and use isolated compute resources. This approach allows multiple data scientists to work on the same pipeline at the same time without over-taxing compute resources, and also makes it easy to use different compute types/sizes for each step.
 
 > [!TIP]
 > ML Pipelines can use run configuration or estimators when training models.
@@ -94,10 +94,11 @@ You can use Azure Machine Learning Compute to distribute the training process ac
 Azure Machine Learning Compute has default limits, such as the number of cores that can be allocated. For more information, see [Manage and request quotas for Azure resources](https://docs.microsoft.com/azure/machine-learning/how-to-manage-quotas).
 
 > [!TIP]
-> Clusters can generally scale upto 100 nodes as long as you have enough quota for the number of cores required. By default clusters are setup with inter-node communication enabled between the nodes of the cluster to support MPI jobs for example. However you can scale your clusters to 1000s of nodes by simply [raising a support ticket](https://portal.azure.com/#blade/Microsoft_Azure_Support/HelpAndSupportBlade/newsupportrequest), and requesting to whitelist your subscription, or workspace, or a specific cluster for disabling inter-node communication. 
->
+> Clusters can generally scale up to 100 nodes as long as you have enough quota for the number of cores required. By default clusters are setup with inter-node communication enabled between the nodes of the cluster to support MPI jobs for example. However you can scale your clusters to 1000s of nodes by simply [raising a support ticket](https://portal.azure.com/#blade/Microsoft_Azure_Support/HelpAndSupportBlade/newsupportrequest), and requesting to whitelist your subscription, or workspace, or a specific cluster for disabling inter-node communication. 
+
+Azure Machine Learning Compute can be reused across runs. The compute can be shared with other users in the workspace and is retained between runs, automatically scaling nodes up or down based on the number of runs submitted, and the max_nodes set on your cluster. The min_nodes setting controls the minimum nodes available.
 
-Azure Machine Learning Compute can be reused across runs. The compute can be shared with other users in the workspace and is retained between runs, automatically scaling nodes up or down based on the number of runs submitted, and the max_nodes set on your cluster.
+[!INCLUDE [min-nodes-note](../../includes/machine-learning-min-nodes.md)]
 
 1. **Create and attach**: To create a persistent Azure Machine Learning Compute resource in Python, specify the **vm_size** and **max_nodes** properties. Azure Machine Learning then uses smart defaults for the other properties. The compute autoscales down to zero nodes when it isn't used.   Dedicated VMs are created to run your jobs as needed.
 
@@ -478,7 +479,7 @@ az ml run submit-hyperdrive -e <experiment> -c <runconfig> --hyperdrive-configur
 
 Note the *arguments* section in runconfig and *parameter space* in HyperDrive config. They contain the command-line arguments to be passed to training script. The value in runconfig stays the same for each iteration, while the range in HyperDrive config is iterated over. Do not specify the same argument in both files.
 
-For more details on these ```az ml``` CLI commands and full set of arguments, see 
+For more details on these ```az ml``` CLI commands, see 
 [the reference documentation](reference-azure-machine-learning-cli.md).
 
 <a id="gitintegration"></a>

includes/machine-learning-min-nodes.md

@@ -0,0 +1,10 @@
+---
+author: Blackmist
+ms.service: machine-learning
+ms.topic: include
+ms.date: 05/08/2020
+ms.author: larryfr
+---
+
+> [!IMPORTANT]
+> To avoid charges when no jobs are running, **set the minimum nodes to 0**. This setting allows Azure Machine Learning to de-allocate the nodes when they aren't in use. Any value larger than 0 will keep that number of nodes running, even if they are not in use.