Submitting tasks/jobs to functions

MLRun batch function objects support a run() method for invoking a job over them. The run method accepts various parameters such as name, handler, params, inputs, schedule, etc. Alternatively you can pass a Task object (see: new_task()) that holds all of the parameters plus the advanced options.

Run/simulate functions locally:

Functions can also run and be debugged locally by using the local runtime or by setting the local=True parameter in the run() method (for batch functions).

Functions can host multiple methods (handlers). You can set the default handler per function. You need to specify which handler you intend to call in the run command.

You can pass parameters (arguments) or data inputs (such as datasets, feature-vectors, models, or files) to the functions through the run method.

  • Inside the function you can access the parameters/inputs by simply adding them as parameters to the function or you can get them from the context object (using get_param() and get_input()).

  • Various data objects (files, tables, models, etc.) are passed to the function as data item objects. You can pass data objects using the inputs dictionary argument, where the dictionary keys match the function’s handler argument names and the MLRun data urls are provided as the values. The data is passed into the function as a DataItem object that handles data movement, tracking and security in an optimal way. Read more about data objects in Data Stores & Data Items.

    run_results = fn.run(params={“label_column”: “label”}, inputs={‘data’: data_url})

MLRun also supports iterative jobs that can run and track multiple child jobs (for hyperparameter tasks, AutoML, etc.). See Hyperparam and Iterative jobs for details and examples.

The run() command returns a run object that you can use to track the job and its results. If you pass the parameter watch=True (default) the run() command blocks until the job completes.

Run object has the following methods/properties:

  • uid() — returns the unique ID.

  • state() — returns the last known state.

  • show() — shows the latest job state and data in a visual widget (with hyperlinks and hints).

  • outputs — returns a dictionary of the run results and artifact paths.

  • logs(watch=True) — returns the latest logs. Use Watch=False to disable the interactive mode in running jobs.

  • artifact(key) — returns an artifact for the provided key (as DataItem object).

  • output(key) — returns a specific result or an artifact path for the provided key.

  • wait_for_completion() — wait for async run to complete

  • refresh() — refresh run state from the db/service

  • to_dict(), to_yaml(), to_json() — converts the run object to a dictionary, YAML, or JSON format (respectively).


You can view the job details, logs, and artifacts in the UI. When you first open the Monitor Jobs tab it displays the last jobs that ran and their data. Click a job name to view its run history, and click a run to view more of the run’s data.


project-jobs-train-artifacts-test_set

See full details and examples in Creating and using functions.