oxirs_vec/joint_embedding_spaces.rs
1//! Joint Embedding Spaces for Cross-Modal Vector Search — thin facade module.
2//!
3//! This module implements advanced joint embedding spaces that enable:
4//! - CLIP-style text-image alignment
5//! - Cross-modal attention mechanisms
6//! - Contrastive learning for alignment
7//! - Multi-modal fusion strategies
8//! - Domain adaptation and transfer learning
9//!
10//! The implementation lives in sibling modules:
11//! - [`crate::joint_embedding_spaces_types`]: configs, structs, enums, type aliases.
12//! - [`crate::joint_embedding_spaces_aligner`]: projectors, attention, scheduler,
13//! domain adapter, and the [`JointEmbeddingSpace`] core.
14//! - [`crate::joint_embedding_spaces_transfer`]: CLIP aligner, augmentation,
15//! curriculum learning, and zero-shot transfer helpers.
16
17pub use crate::joint_embedding_spaces_aligner::*;
18pub use crate::joint_embedding_spaces_transfer::*;
19pub use crate::joint_embedding_spaces_types::*;