pelican.layers.PELICANBlock

class pelican.layers.PELICANBlock(hidden_channels, increase_hidden_channels=1.0, activation='leaky_relu', dropout_prob=None, **kwargs)[source]

Bases: Module

PELICAN edge-to-edge aggregation block.

A single PELICAN block consisting of a feedforward network and edge-to-edge aggregation.

Parameters:

  • hidden_channels (int) – Number of hidden channels.

  • increase_hidden_channels (float) – Factor to increase hidden channels in the feedforward network, by default 1.0.

  • activation (str) – Activation function to use (‘gelu’, ‘relu’, ‘leaky_relu’, ‘tanh’, ‘sigmoid’, ‘silu’), by default ‘leaky_relu’.

  • dropout_prob (float) – Dropout probability, by default None.

  • **kwargs – Additional keyword arguments to pass to the aggregator.

forward(x, edge_index, batch, **kwargs)[source]

Forward pass of the PELICAN block.

Parameters:

  • x (torch.Tensor) – Edge-level features of shape (E, C).

  • edge_index (torch.Tensor) – Edge index tensor of shape (2, E).

  • batch (torch.Tensor) – Batch tensor of shape (N).

  • **kwargs

Returns:

x – Updated edge-level features of shape (E, C).

Return type:

torch.Tensor