oxicuda-gnn 0.2.0

Graph Neural Network primitives for OxiCUDA: sparse graph representations (CSR/COO/heterogeneous), message passing framework, GCN/GAT/GraphSAGE/GIN layers, global and hierarchical graph pooling — pure Rust, zero CUDA SDK dependency.
Documentation 
1
2
3
4
5
6
7
8
9

//! Graph sampling algorithms for mini-batch GNN training.

pub mod cluster_gcn;
pub mod graphsaint;
pub mod neighbor_sample;

pub use cluster_gcn::{BatchSubgraph, ClusterGcn, Partition};
pub use graphsaint::{GraphSaint, SaintNorm, SaintSampler, SaintSubgraph};
pub use neighbor_sample::{GnnRng, NeighborSampleConfig, NeighborSampler, SampledSubgraph};