Struct FinalReport 

pub struct FinalReport {
    pub direction: Direction,
    pub best: Option<TrialNumber>,
    pub entries: Vec<ReevalEntry>,
}

The outcome of the final test-seed re-evaluation stage (Study::reevaluate).

It carries the study’s honest, held-out-seed answer — best, the test-seed-ranked best trial — alongside the per-configuration tune-vs-test comparison that exposes overfitting.

Fields

direction: Direction

The study’s single objective direction (the stage is single-objective).

best: Option<TrialNumber>

The test-seed-ranked best trial, by its per-study number; None if nothing was re-evaluated or every candidate was ineligible.

entries: Vec<ReevalEntry>

One entry per re-evaluated configuration, in tune-ranked order.

Implementations

impl FinalReport

pub fn best_entry(&self) -> Option<&ReevalEntry>

The re-evaluation entry for the test-ranked best trial, if there is one.

pub fn best_changed_under_test(&self) -> bool

Whether the test-ranked best differs from the tune-ranked best — the signal that tuning overfit its seeds.

true when re-ranking on held-out seeds moved a different configuration to the top than the one the tune aggregate preferred.

Trait Implementations

impl Clone for FinalReport

fn clone(&self) -> FinalReport

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for FinalReport

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for FinalReport

fn eq(&self, other: &FinalReport) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for FinalReport