Azure ML Experiments and Runs

An experiment is a grouping of many runs from a specified script. It always belongs to a workspace. When we submit a run, we provide an experiment name. Information for the run is stored under that experiment. If the name doesn't exist when we submit an experiment, a new experiment is automatically created.

A run is a single execution of a training script. An experiment will typically contain multiple runs.

A run has the following characteristics

• Metrics
• Child Runs
• Outputs
• Logs

A run also has metrics and parameters associated with it

A run also has outputs associated with it

Code

Create the workspace

from azureml.core import Workspace
from azureml.core import Workspace
from azureml.core.authentication import InteractiveLoginAuthentication

sid = '<your-subscription-id>'
forced_interactive_auth = InteractiveLoginAuthentication(tenant_id="<your-tenant-id>", force=True)

ws = Workspace.create(name='azureml_workspace',
            subscription_id= sid, 
            resource_group='rgazureml',
            create_resource_group = True,
            location='centralus'
            )

Create an experiment and a run

from azureml.core import Experiment

# create an experiment
exp = Experiment(workspace=ws, name='trial_exp')

# start a run
run = exp.start_logging()

# log a number
run.log('trial', 30)

# log a list (Fibonacci numbers)
run.log_list('my list', [1, 1, 2, 3, 5, 8, 13, 21, 34, 55]) 

# finish the run
run.complete()

print('Finished logging')

Create another run

# start a run
run = exp.start_logging()

# log a number
run.log('trial2', 35)

# log a list 
run.log_list('my list2', [1, 1, 2, 2, 5, 5, 13, 13, 13, 13]) 

# finish the run
run.complete()

References

• Azure Machine Learning documentation