Struct ConfidenceCrossEncoder 

pub struct ConfidenceCrossEncoder { /* private fields */ }

Confidence-based cross-encoder implementation

This implementation uses semantic similarity and confidence metrics to rerank results. For production use with actual transformer models, consider using ONNXCrossEncoder or APICrossEncoder implementations.

Implementations

impl ConfidenceCrossEncoder

pub fn new(config: CrossEncoderConfig) -> Self

Create a new confidence-based cross-encoder

Trait Implementations

impl CrossEncoder for ConfidenceCrossEncoder

fn rerank<'life0, 'life1, 'async_trait>( &'life0 self, query: &'life1 str, candidates: Vec<SearchResult>, ) -> Pin<Box<dyn Future<Output = Result<Vec<RankedResult>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Rerank a list of search results based on relevance to query

fn score_pair<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, query: &'life1 str, document: &'life2 str, ) -> Pin<Box<dyn Future<Output = Result<f32>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Score a single query-document pair

fn score_batch<'life0, 'async_trait>( &'life0 self, pairs: Vec<(String, String)>, ) -> Pin<Box<dyn Future<Output = Result<Vec<f32>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Batch score multiple query-document pairs