Struct CrossModalEncoder 

pub struct CrossModalEncoder { /* private fields */ }

Cross-modal embedding encoder that handles multiple modalities

Implementations

impl CrossModalEncoder

pub fn new( config: CrossModalConfig, text_encoder: Box<dyn TextEncoder>, image_encoder: Box<dyn ImageEncoder>, audio_encoder: Box<dyn AudioEncoder>, video_encoder: Box<dyn VideoEncoder>, graph_encoder: Box<dyn GraphEncoder>, ) -> Self

pub fn encode(&self, content: &MultiModalContent) -> Result<Vector>

Encode multi-modal content into a joint embedding space

pub fn cross_modal_similarity( &self, content1: &MultiModalContent, content2: &MultiModalContent, ) -> Result<f32>

Calculate cross-modal similarity between two multi-modal contents

pub fn find_cross_modal_matches( &self, query_content: &MultiModalContent, candidates: &[MultiModalContent], top_k: usize, ) -> Result<Vec<(usize, f32)>>

Find cross-modal matches (e.g., images that match text descriptions)

pub fn align_modalities( &mut self, paired_data: &[(MultiModalContent, MultiModalContent)], ) -> Result<()>

Align embeddings across modalities using contrastive learning

pub fn get_alignment_stats(&self) -> (usize, f32)

Get alignment statistics