torch.as_strided

function as_strided(input: Tensor, size: readonly number[], stride: readonly number[], options?: AsStridedOptions): Tensor

function as_strided(input: Tensor, size: readonly number[], stride: readonly number[], storage_offset: number, options?: AsStridedOptions): Tensor

Creates a view of the input tensor with the given size and stride.

This allows creating views that overlap or have gaps in the underlying storage. The resulting tensor shares storage with the input tensor.

Parameters

Tensor– A new view of the input tensor

const x = torch.arange(9);  // [0, 1, 2, 3, 4, 5, 6, 7, 8]
torch.as_strided(x, [2, 2], [3, 1]);  // [[0, 1], [3, 4]]

function as_strided(input: Tensor, size: readonly number[], stride: readonly number[], options?: AsStridedOptions): Tensor

function as_strided(input: Tensor, size: readonly number[], stride: readonly number[], storage_offset: number, options?: AsStridedOptions): Tensor

Creates a view of the input tensor with the given size and stride.

This allows creating views that overlap or have gaps in the underlying storage. The resulting tensor shares storage with the input tensor.

Tensor– A new view of the input tensor

const x = torch.arange(9);  // [0, 1, 2, 3, 4, 5, 6, 7, 8]
torch.as_strided(x, [2, 2], [3, 1]);  // [[0, 1], [3, 4]]