AttentionEdgeWeighting

class AttentionEdgeWeighting(message_dim: int, num_heads: int = 8, dropout: float = 0.1)

Bases: EdgeWeighting

Message weighting by attention.

Initialize the module.

Parameters:

• message_dim (int) – >0 the message dimension. has to be divisible by num_heads .. todo:: change to multiplicative instead of divisive to make this easier to use

• num_heads (int) – >0 the number of attention heads

• dropout (float) – the attention dropout

Raises:

ValueError – If message_dim is not divisible by num_heads

Attributes Summary

needs_message

whether the edge weighting needs access to the message

Methods Summary

forward(source, target[, message, x_e])

Compute edge weights.

Attributes Documentation

needs_message: ClassVar[bool] = True

whether the edge weighting needs access to the message

Methods Documentation

forward(source: Tensor, target: Tensor, message: Tensor | None = None, x_e: Tensor | None = None) → Tensor

Compute edge weights.

Parameters:

• source (Tensor) – shape: (num_edges,) The source indices.

• target (Tensor) – shape: (num_edges,) The target indices.

• message (Tensor | None) – shape (num_edges, dim) Actual messages to weight

• x_e (Tensor | None) – shape (num_nodes, dim) Node states up to the weighting point

Returns:

shape: (num_edges, dim) Messages weighted with the edge weights.

Return type:

Tensor