Events
Mar 17, 9 PM - Mar 21, 10 AM
Join the meetup series to build scalable AI solutions based on real-world use cases with fellow developers and experts.
Register nowThis browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
The following is a guide to help you learn more about resources used to train models in Azure with Model Builder.
An Azure Machine Learning experiment is a resource that needs to be created before running Model Builder training on Azure.
The experiment encapsulates the configuration and results for one or more machine learning training runs. Experiments belong to a specific workspace. The first time an experiment is created, its name is registered in the workspace. Any subsequent runs - if the same experiment name is used - are logged as part of the same experiment. Otherwise, a new experiment is created.
A workspace is an Azure Machine Learning resource that provides a central place for all Azure Machine Learning resources and artifacts created as part of training run.
To create an Azure Machine Learning workspace, the following are required:
An Azure Machine Learning compute is a cloud-based Linux VM used for training.
To create an Azure Machine Learning compute, the following values are required:
Name: A name for your compute between 2-16 characters. Names may only contain alphanumeric characters and hyphens.
Compute size.
Model Builder can use one of the following GPU-optimized compute types:
Size | vCPU | Memory: GiB | Temp storage (SSD) GiB | GPU | GPU memory: GiB | Max data disks | Max NICs |
---|---|---|---|---|---|---|---|
Standard_NC12 | 12 | 112 | 680 | 2 | 24 | 48 | 2 |
Standard_NC24 | 24 | 224 | 1440 | 4 | 48 | 64 | 4 |
Visit the NC-series Linux VM documentation for more details on GPU optimized compute types.
Compute priority.
Training on Azure is only available for the Model Builder image classification scenario. The algorithm used to train these models is a Deep Neural Network based on the ResNet50 architecture. The training process takes some time and the amount of time may vary depending on the size of compute selected as well as amount of data. You can track the progress of your runs by selecting the "Monitor current run in Azure portal" link in Visual Studio.
Once training is complete, two projects are added to your solution with the following suffixes:
ConsoleApp: A C# console app that provides starter code to build the prediction pipeline and make predictions.
Model: A C# .NET Standard app that contains the data models that define the schema of input and output model data as well as the following assets:
bestModelMap.json
file.The ModelInput
and ModelOutput
classes in the Model project define the schema of the model's expected input and output respectively.
In an image classification scenario, the ModelInput
contains two columns:
ImageSource
: The string path of the image location.Label
: The actual category the image belongs to. Label
is only used as an input when training and does not need to be provided when making predictions.The ModelOutput
contains two columns:
Prediction
: The image's predicted category.Score
: The list of probabilities for all categories (the highest belongs to the Prediction
).If an error occurs during Azure Machine Learning compute creation, the compute resource may still exist, in an errored state. If you try to re-create the compute resource with the same name, the operation fails. To fix this error, either:
.NET feedback
.NET is an open source project. Select a link to provide feedback:
Events
Mar 17, 9 PM - Mar 21, 10 AM
Join the meetup series to build scalable AI solutions based on real-world use cases with fellow developers and experts.
Register now