Use compute targets

Completed

After you've created or attached compute targets in your workspace, you can use them to run specific workloads; such as experiments.

To use a particular compute target, you can specify it in the appropriate parameter for an experiment run configuration or estimator. For example, the following code configures an estimator to use the compute target named aml-cluster:

from azureml.core import Environment, ScriptRunConfig

compute_name = 'aml-cluster'

training_env = Environment.get(workspace=ws, name='training_environment')

script_config = ScriptRunConfig(source_directory='my_dir',
                                script='script.py',
                                environment=training_env,
                                compute_target=compute_name)

When an experiment is submitted, the run will be queued while the aml-cluster compute target is started and the specified environment created on it, and then the run will be processed on the compute environment.

Instead of specifying the name of the compute target, you can specify a ComputeTarget object, like this:

from azureml.core import Environment, ScriptRunConfig
from azureml.core.compute import ComputeTarget

compute_name = "aml-cluster"

training_cluster = ComputeTarget(workspace=ws, name=compute_name)

training_env = Environment.get(workspace=ws, name='training_environment')

script_config = ScriptRunConfig(source_directory='my_dir',
                                script='script.py',
                                environment=training_env,
                                compute_target=training_cluster)