ReinforcementLearningRun Class

A run class to handle and monitor Reinforcement Learning Runs associated with an experiment and an individual run ID.

Class ReinforcementLearningRun constructor.

Inheritance
ReinforcementLearningRun

Constructor

ReinforcementLearningRun(experiment, run_id, directory=None, _run_config=None, **kwargs)

Parameters

Name Description
experiment
Required

The experiment object.

run_id
Required
str

The run id.

directory
str

The source directory.

Default value: None
_run_config

The reinforcement learning configuration.

Default value: None
kwargs
Required

Remarks

The Azure Machine Learning SDK provides you with a series of interconnected classes, that are designed to help you train and compare machine learning models that are related by the shared problem that they are solving.

An Experiment acts as a logical container for these training runs. A ReinforcementLearningConfiguration object is used to codify the information necessary to submit a training run in a Reinforcement Learning experiment. Which can then be submitted through the experiment. Please refer to the documentation in ReinforcementLearningConfiguration to see an example of this process.

Once the ReinforcementLearningConfiguration is submitted, a ReinforcementLearningRun object is returned.

A ReinforcementLearningRun object gives you programmatic access to information about the associated Reinforcement Learning run. Some examples include retrieving the logs corresponding to a run, canceling or completing a run if it's still in progress, cleaning up the artifacts of a completed run, and waiting for completion of a run currently in progress.

Methods

complete

Complete the ongoing run

complete

Complete the ongoing run

complete()

Remarks

An example to complete run is as follows:


   run = experiment.submit(config=ReinforcementLearningRunConfig)
   run.complete()

Attributes

RUN_TYPE

RUN_TYPE = 'reinforcementlearning'