choose_trial_to_run

PopulationBasedTraining.choose_trial_to_run(tune_controller: TuneController) → Trial | None

Ensures all trials get fair share of time (as defined by time_attr).

This enables the PBT scheduler to support a greater number of concurrent trials than can fit in the cluster at any given time.

Parameters:

tune_controller (TuneController) – The controller running the experiment.

Returns:

The PENDING or PAUSED trial with the lowest time_attr value as of its last perturbation interval, or None if there’s none. In synchronous mode, only trials that haven’t reached the next perturbation point are candidates.

Return type:

Trial | None