Watch the video below to learn how to effectively deploy and manage your virtual machines and compute resources within Azure ML Studio.
Effective compute management is crucial for optimising both performance and costs in your machine learning projects. This video highlights the three main types of compute resource avaiable in Azure ML Studio:
1 Serverless Compute
Automatically managed compute that scales on-demand without requiring manual configuration. Ideal for quick experiments and prototyping where you need immediate access to computational resources.
2 Single Compute Instances
Dedicated virtual machines that you can customise and control directly. Perfect for development work, debugging, and scenarios requiring specific software configurations or persistent environments.
3 Compute Clusters
Scalable clusters that can dynamically adjust their size based on workload demands. Essential for large-scale training jobs, hyperparameter tuning, and distributed computing tasks.
Cost Management Tip: Remember to stop or delete compute instances when not in use to avoid unnecessary charges. Serverless compute automatically manages this for you, whilst single instances and clusters require manual oversight.