torch_tensor_from_buffer {torch} | R Documentation |
Creates a tensor from a buffer of memory
Description
It creates a tensor without taking ownership of the memory it points to.
You must call clone
if you want to copy the memory over a new tensor.
Usage
torch_tensor_from_buffer(buffer, shape, dtype = "float")
buffer_from_torch_tensor(tensor)
Arguments
buffer |
An R atomic object containing the data in a contiguous array. |
shape |
The shape of the resulting tensor. |
dtype |
A torch data type for the tresulting tensor. |
tensor |
Tensor object that will be converted into a buffer. |
Functions
-
buffer_from_torch_tensor()
: Creates a raw vector containing the tensor data. Causes a data copy.
[Package torch version 0.13.0 Index]