nip.utils.torch.PairInvariantizer

class nip.utils.torch.PairInvariantizer(pair_dim: int = 0)

Transform the input to be invariant to the order of the graphs in a pair.

Works by taking the mean of the pair and half the absolute difference between the graphs.

Parameters:

pair_dim (int, default=0) – The graph pair dimension.

Methods Summary

__init__([pair_dim])	Initialize internal Module state, shared by both nn.Module and ScriptModule.
forward(x)	Transform the input to be invariant to the order of the graphs in a pair.

Attributes

T_destination
call_super_init
dump_patches
training

Methods

__init__(pair_dim: int = 0)

Initialize internal Module state, shared by both nn.Module and ScriptModule.

forward(x: Tensor) → Tensor

Transform the input to be invariant to the order of the graphs in a pair.

Parameters:

x (Tensor) – The input tensor.

Returns:

transformed_x (Tensor) – The input tensor transformed to be invariant to the order of the graphs in a pair