Struct MedianPruner 

pub struct MedianPruner {
    pub n_warmup_steps: usize,
    pub min_trials: usize,
}

Prune if current value is below the median of completed trials at the same step.

Fields

n_warmup_steps: usize

Don’t prune before this many steps.

min_trials: usize

Minimum completed trials needed before pruning starts.

Implementations

impl MedianPruner

pub fn new(n_warmup_steps: usize) -> Self

pub fn with_min_trials(self, min_trials: usize) -> Self

Trait Implementations

impl Pruner for MedianPruner

fn should_prune( &self, metric_name: &str, current_value: f64, step: usize, history: &[TrialMetricHistory], ) -> Option<String>

Decide whether to prune given the current trial’s metrics and the history of completed trials’ metrics at the same step.