ray.tune.schedulers.HyperBandForBOHB#

class ray.tune.schedulers.HyperBandForBOHB(time_attr: str = 'training_iteration', metric: Optional[str] = None, mode: Optional[str] = None, max_t: int = 81, reduction_factor: float = 3, stop_last_trials: bool = True)[source]#

Bases: ray.tune.schedulers.hyperband.HyperBandScheduler

Extends HyperBand early stopping algorithm for BOHB.

This implementation removes the HyperBandScheduler pipelining. This class introduces key changes:

1. Trials are now placed so that the bracket with the largest size is filled first.

2. Trials will be paused even if the bracket is not filled. This allows BOHB to insert new trials into the training.

See ray.tune.schedulers.HyperBandScheduler for parameter docstring.

PublicAPI: This API is stable across Ray releases.

Methods

choose_trial_to_run(tune_controller[, ...])

Fair scheduling within iteration by completion percentage.

debug_string()

This provides a progress notification for the algorithm.

on_trial_add(tune_controller, trial)

Adds new trial.

on_trial_complete(tune_controller, trial, result)

Cleans up trial info from bracket if trial completed early.

on_trial_error(tune_controller, trial)

Cleans up trial info from bracket if trial errored early.

on_trial_remove(tune_controller, trial)

Notification when trial terminates.

on_trial_result(tune_controller, trial, result)

If bracket is finished, all trials will be stopped.

restore(checkpoint_path)

Restore trial scheduler from checkpoint.

save(checkpoint_path)

Save trial scheduler to a checkpoint

Attributes

CONTINUE

Status for continuing trial execution

NOOP

PAUSE

Status for pausing trial execution

STOP

Status for stopping trial execution

metric

supports_buffered_results