OrthogonalityRegularizer

class OrthogonalityRegularizer(*, weight: float = 1.0, apply_only_once: bool = True, epsilon: float = 1e-05, **kwargs)

Bases: Regularizer

A regularizer for the soft orthogonality constraints from [wang2014].

Initialize the regularizer.

Parameters:

• weight (Tensor) – The relative weight of the regularization

• apply_only_once (bool) – Should the regularization be applied more than once after reset?

• epsilon (float) – a small value used to check for approximate orthogonality

• kwargs – additional keyword-based parameters passed to Regularizer.__init__()

Methods Summary

forward(x)	Compute the regularization term for one tensor.
update(*tensors)	Update the regularization term based on passed tensors.

Methods Documentation

forward(x: Tensor) → Tensor

Compute the regularization term for one tensor.

Parameters:

x (Tensor)

Return type:

Tensor

update(*tensors: Tensor) → None

Update the regularization term based on passed tensors.

Parameters:

tensors (Tensor)

Return type:

None