torch.roll()

torch.roll(input, shifts, dims=None) -> Tensor

Definition

Scrolls the tensor input along the given dimension.Elements beyond the last position are reintroduced at the first position.If dims is None, the tensor will be flattened before rolling and then restored to its original shape.

Parameters

• input (Tensor) - input tensor
• shifts (int or tuple of python: ints) - The number of positions to shift the elements of the tensor.If shifts is a tuple, dims must be a tuple of the same size, with each dimension rolling the corresponding value
• dims (int or tuple of python: ints) - scroll along the dimension specified by dims

Example

import torchx = torch.tensor([1, 2, 3, 4, 5, 6, 7, 8]).view(4, 2)print(x)# tensor([[1, 2],# [3, 4],# [5, 6],# [7, 8]])x1 = torch.roll(x, 1)print(x1)# tensor([[8, 1],#         [twenty three],# [4, 5],# [6, 7]])x2 = torch.roll(x, 1, 0)print(x2)# tensor([[7, 8],# [1, 2],# [3, 4],# [5, 6]])x3 = torch.roll(x, -1, 0)print(x3)# tensor([[3, 4],# [5, 6],# [7, 8],# [1, 2]])x4 = torch.roll(x, shifts=(2, 1), dims=(0, 1))print(x4)# tensor([[6, 5],# [8, 7],#         [twenty one],# [4, 3]])

Reference

torch.roll — PyTorch 1.12 documentation