neuralop.layers.normalization_layers.InstanceNorm

class neuralop.layers.normalization_layers.InstanceNorm(**kwargs)[source]

Dimension-agnostic instance normalization layer for neural operators.

InstanceNorm normalizes each sample in the batch independently, computing mean and variance across spatial dimensions for each sample and channel separately. This is useful when the statistical properties of each sample are distinct and should be treated separately.

Parameters:

**kwargsdict, optional

Additional parameters to pass to torch.nn.functional.instance_norm(). Common parameters include: - eps : float, optional

Small value added to the denominator for numerical stability. Default is 1e-5.

• momentumfloat, optional

  Value used for the running_mean and running_var computation. Default is 0.1.

• use_input_statsbool, optional

  If True, use input statistics. Default is True.

• weighttorch.Tensor, optional

  Weight tensor for affine transformation. If None, no scaling applied.

• biastorch.Tensor, optional

  Bias tensor for affine transformation. If None, no bias applied.

Methods

forward(x)

Apply instance normalization to the input tensor.

forward(x)[source]

Apply instance normalization to the input tensor.