ray.tune.stopper.ExperimentPlateauStopper
ray.tune.stopper.ExperimentPlateauStopper#
- class ray.tune.stopper.ExperimentPlateauStopper(metric: str, std: float = 0.001, top: int = 10, mode: str = 'min', patience: int = 0)[source]#
Bases:
ray.tune.stopper.stopper.Stopper
Early stop the experiment when a metric plateaued across trials.
Stops the entire experiment when the metric has plateaued for more than the given amount of iterations specified in the patience parameter.
- Parameters
metric β The metric to be monitored.
std β The minimal standard deviation after which the tuning process has to stop.
top β The number of best models to consider.
mode β The mode to select the top results. Can either be βminβ or βmaxβ.
patience β Number of epochs to wait for a change in the top models.
- Raises
ValueError β If the mode parameter is not βminβ nor βmaxβ.
ValueError β If the top parameter is not an integer greater than 1.
ValueError β If the standard deviation parameter is not a strictly positive float.
ValueError β If the patience parameter is not a strictly positive integer.
PublicAPI: This API is stable across Ray releases.