Struct GATConv 

pub struct GATConv { /* private fields */ }

Graph Attention Network layer (Velickovic et al., 2018).

Uses attention mechanism to weight neighbor contributions:

α_ij = softmax_j(LeakyReLU(a^T [W*h_i || W*h_j]))
h_i' = σ(Σ_j α_ij * W * h_j)

Example

use aprender::gnn::GATConv;

let gat = GATConv::new(16, 32, 4);  // 16 in → 32 out, 4 attention heads

Implementations

impl GATConv

pub fn new(in_features: usize, out_features: usize, num_heads: usize) -> GATConv

Create a new GAT convolutional layer.

Arguments

• in_features - Input feature dimension
• out_features - Output feature dimension per attention head
• num_heads - Number of attention heads

pub fn num_heads(&self) -> usize

Get number of attention heads.

pub fn out_features(&self) -> usize

Get output dimension per head.

pub fn total_out_features(&self) -> usize

Total output dimension (out_features * num_heads).

Trait Implementations

impl Debug for GATConv

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl GNNModule for GATConv

fn forward_gnn(&self, x: &Tensor, edge_index: &[(usize, usize)]) -> Tensor

Forward pass with graph structure.

impl Module for GATConv

fn forward(&self, _input: &Tensor) -> Tensor

Perform forward computation.

fn parameters(&self) -> Vec<&Tensor>

Get references to all learnable parameters.

fn parameters_mut(&mut self) -> Vec<&mut Tensor>

Get mutable references to all learnable parameters.

fn named_parameters(&self) -> Vec<(String, &Tensor)>

Get (name, parameter) pairs for all learnable parameters.

fn named_parameters_mut(&mut self) -> Vec<(String, &mut Tensor)>

Get (name, mutable parameter) pairs for all learnable parameters.

fn set_training(&mut self, training: bool)

Set training mode recursively, propagating into child modules.

fn refresh_caches(&mut self)

Refresh any cached computations after parameters have been modified.

fn train(&mut self)

Set the module to training mode.

fn eval(&mut self)

Set the module to evaluation mode.

fn training(&self) -> bool

Check if the module is in training mode.

fn zero_grad(&mut self)

Zero out gradients for all parameters.

fn num_parameters(&self) -> usize

Get the number of learnable parameters.