Struct DecontaminationReport 

pub struct DecontaminationReport {
    pub ngram_size: usize,
    pub threshold: f64,
    pub total_samples: usize,
    pub evaluated_samples: usize,
    pub evaluated_references: usize,
    pub contaminated_count: usize,
    pub contamination_rate: f64,
    pub flagged: Vec<ContaminationResult>,
}

Summary report of decontamination check.

Fields

ngram_size: usize

N-gram size used

threshold: f64

Overlap threshold used

total_samples: usize

Total training samples checked

evaluated_samples: usize

Training samples that yielded at least one n-gram and were therefore actually compared against the reference set.

Samples shorter than ngram_size (and every sample when ngram_size is 0) produce no n-grams and are silently un-checkable. When this is 0 but total_samples is not, the check was vacuous: a clean contamination_rate of 0.0 proves nothing.

evaluated_references: usize

Reference samples that yielded at least one n-gram. Zero here also makes the check vacuous — nothing could ever be matched against.

contaminated_count: usize

Number of contaminated samples

contamination_rate: f64

Contamination rate (0.0 to 1.0)

flagged: Vec<ContaminationResult>

Per-sample results (only contaminated samples included)

Implementations

impl DecontaminationReport

pub fn is_vacuous(&self) -> bool

Whether the configuration made the check meaningless.

True when there was input to check but nothing could be compared — e.g. ngram_size exceeds the length of every sample, or is 0. A caller enforcing a contamination gate must treat this as a failure, not as a pass.

Trait Implementations

impl Clone for DecontaminationReport

fn clone(&self) -> DecontaminationReport

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DecontaminationReport

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

Formats the value using the given formatter.