See the Glossary for the meaning of the acronyms used in this guide.


run_mlflow_task(workflow_uri: str, entry_point: str, experiment_id: str, conda_env: str = 'base', entry_point_kwargs: Optional[str] = None, s3: Optional[botocore.client.BaseClient] = None) subprocess.CompletedProcess[source]#
run_task_engine_task(experiment_id: int, experiment_desc: Mapping[str, Any], global_parameters: MutableMapping[str, Any], s3: Optional[botocore.client.BaseClient] = None)[source]#

Run an experiment via the task engine.

  • experiment_id – The ID of the experiment to use for this run

  • experiment_desc – A declarative experiment description, as a mapping

  • global_parameters – Global parameters for this run, as a mapping from parameter name to value

  • s3 – An optional boto3 S3 client object to use. If not given, construct one according to environment variables. This argument is not normally used, but useful in unit tests when you need a specially configured object with stubbed responses.