torch.clone¶

torch.clone(input, *, memory_format=torch.preserve_format) → Tensor¶

Returns a copy of input.

Note

This function is differentiable, so gradients will flow back from the result of this operation to input. To create a tensor without an autograd relationship to input see detach().

Parameters

input (Tensor) – the input tensor.

Keyword Arguments

memory_format (torch.memory_format, optional) – the desired memory format of returned tensor. Default: torch.preserve_format.