edits

Author: GitHubber17

articles/machine-learning/how-to-attach-kubernetes-anywhere.md

@@ -61,10 +61,10 @@ With Arc Kubernetes cluster, you can build, train, and deploy models in any on-p
 
 | Usage pattern | Location of data | Goals and requirements | Scenario configuration |
 | --- | --- | --- | --- |
-| Train model in cloud, deploy model on-premises | Cloud | - _Use cloud compute to support elastic compute needs or special hardware such as a GPU._ <br> - _Model deployment must be on-premises for security, compliance, or latency requirements._ | - Azure-managed compute in cloud <br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
-| Train model on-premises and cloud, deploy to both cloud and on-premises | Cloud | - _Combine on-premises investments with cloud scalability._ <br> - _Bring cloud and on-premises compute under single pane of glass._ <br> - _Access single source of truth for data in cloud and replicate on-premises (lazily on usage or proactively)._ <br> - _Enable cloud compute primary usage when on-premises resources aren't available (in use or in maintenance) or don't meet specific hardware requirements (GPU)._ | - Azure-managed compute in cloud. br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
-| Train model on-premises, deploy model in cloud | On-premises | - _Store data on-premises to meet data-residency requirements._ <br> - _Deploy model in the cloud for global-service access or to enable compute elasticity for scale and throughput._ | - Azure-managed compute in cloud <br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
-| Bring your own AKS in Azure | Cloud | - _Gain more security and controls._ <br> - _Establish all private IP machine learning to prevent data exfiltration._ | - AKS cluster behind an Azure virtual network <br> - Private endpoints in the same virtual network for Azure Machine Learning workspace and associated resources <br> Fully automated machine learning operations |
+| Train model in cloud, deploy model on-premises | Cloud | _Use cloud compute to support elastic compute needs or special hardware such as a GPU._ <br><br> _Model deployment must be on-premises for security, compliance, or latency requirements._ | - Azure-managed compute in cloud <br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
+| Train model on-premises and cloud, deploy to both cloud and on-premises | Cloud | _Combine on-premises investments with cloud scalability._ <br><br> _Bring cloud and on-premises compute under single pane of glass._ <br><br> _Access single source of truth for data in cloud and replicate on-premises (lazily on usage or proactively)._ <br><br> _Enable cloud compute primary usage when on-premises resources aren't available (in use or in maintenance) or don't meet specific hardware requirements (GPU)._ | - Azure-managed compute in cloud. <br> Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
+| Train model on-premises, deploy model in cloud | On-premises | _Store data on-premises to meet data-residency requirements._ <br><br> _Deploy model in the cloud for global-service access or to enable compute elasticity for scale and throughput._ | - Azure-managed compute in cloud <br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations in hybrid mode, including training and model deployment steps that transition seamlessly between cloud and on-premises <br> - Repeatable, all assets properly tracked, model retrained as needed, deployment updated automatically after retraining |
+| Bring your own AKS in Azure | Cloud | _Gain more security and controls._ <br><br> _Establish all private IP machine learning to prevent data exfiltration._ | - AKS cluster behind an Azure virtual network <br> - Private endpoints in the same virtual network for Azure Machine Learning workspace and associated resources <br> Fully automated machine learning operations |
 | Full machine learning lifecycle on-premises | On-premises | _Secure sensitive data or proprietary IP, such as machine learning models, code, and scripts._ | - Outbound proxy server connection on-premises <br> - Azure ExpressRoute and Azure Arc private link to Azure resources <br> - Customer-managed Kubernetes on-premises <br> - Fully automated machine learning operations |
 
 ### Limitations for Kubernetes compute target