Tensor.to_device

Tensor.to_device(device: str)[source]

Return a copy of this element with the same values stored on a different computational device.

Parameters

device :

Identifier of the desired storage location. Which ones are supported depends on the array backend (impl). Always allowed is 'cpu', but GPU alternatives like 'cuda:0' can offer better performance if available.

Returns

newelemTensor

Version of this element with its data array on the desired device.