mlx_graphs.nn.GCNConv

mlx_graphs.nn.GCNConv#

class mlx_graphs.nn.GCNConv(node_features_dim: int, out_features_dim: int, bias: bool = True, add_self_loops: bool = False, **kwargs)[source]#

Bases: MessagePassing

Applies a GCN convolution over input node features.

Parameters:

  • node_features_dim (int) – size of input node features

  • out_features_dim (int) – size of output node embeddings

  • bias (bool) – whether to use bias in the node projection

  • add_self_loops (bool) – whether to add a self-loop for each node

__call__(edge_index: mlx.core.array, node_features: mlx.core.array, edge_weights: mlx.core.array | None = None, normalize: bool = True, **kwargs: Any) → mlx.core.array[source]#

Call self as a function.

Return type:

array

Methods

Contents