ray.tune.schedulers.TrialScheduler#

class ray.tune.schedulers.TrialScheduler[source]#

Bases: object

Interface for implementing a Trial Scheduler class.

Note to Tune developers: If a new scheduler is added, please update air/_internal/usage.py.

DeveloperAPI: This API may change across minor Ray releases.

Methods

choose_trial_to_run(tune_controller)

Called to choose a new trial to run.

debug_string()

Returns a human readable message for printing to the console.

on_trial_add(tune_controller, trial)

Called when a new trial is added to the trial runner.

on_trial_complete(tune_controller, trial, result)

Notification for the completion of trial.

on_trial_error(tune_controller, trial)

Notification for the error of trial.

on_trial_remove(tune_controller, trial)

Called to remove trial.

on_trial_result(tune_controller, trial, result)

Called on each intermediate result returned by a trial.

restore(checkpoint_path)

Restore trial scheduler from checkpoint.

save(checkpoint_path)

Save trial scheduler to a checkpoint

set_search_properties(metric, mode, **spec)

Pass search properties to scheduler.

Attributes

CONTINUE

Status for continuing trial execution

NOOP

PAUSE

Status for pausing trial execution

STOP

Status for stopping trial execution

metric

supports_buffered_results