bartorch.normalize

bartorch.normalize()

Divide by the norm over axes, separately for each index of the other axes.

Parameters:

• input (torch.Tensor) – Array to normalize.

• axes (int or tuple of int) – Axes the norm is taken over.

• l1 (bool, default=False) – Use the l1 norm rather than the l2 norm.

Returns:

The shape of input.

Return type:

torch.Tensor