ruvector-attention 2.1.0

Attention mechanisms for ruvector - geometric, graph, and sparse attention
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

//! PDE-Based Attention
//!
//! Continuous-time attention using partial differential equations.
//!
//! ## Key Concepts
//!
//! 1. **Diffusion Smoothing**: Heat equation on attention graph
//! 2. **Graph Laplacian**: L = D - W for key similarity
//! 3. **Time Evolution**: x_{t+dt} = x_t - dt * L * x_t
//!
//! ## Interpretation
//!
//! - Attention as continuous information flow
//! - Smoothing removes noise while preserving structure
//! - Multi-scale attention via different diffusion times

mod diffusion;
mod laplacian;

pub use diffusion::{DiffusionAttention, DiffusionConfig};
pub use laplacian::{GraphLaplacian, LaplacianType};

#[cfg(test)]
mod tests {
    #[test]
    fn test_module_exists() {
        assert!(true);
    }
}