Query regarding PyTorch tensors and NumPy

[VijayV28]

Hey,

It was mentioned that when a Numpy array is converted to a Tensor, the tensor won't change even if the numpy is reassigned. Have a look at the code below:

Converting a Numpy array to tensor

array_x = np.arange(1,8)
tensor_x = torch.from_numpy(array_x)
array_x, tensor_x

I tried reassigning a number in the array as shown below:

array_x[0] = 9
array_x, tensor_x

The above change was reflected in the tensor:
Output

However, when I tried reassigning using the method given in the video, the change was not shown in the tensor:

array_x = array_x + 1
array_x, tensor_x

Output

Can anyone please explain why is this happening?

[AlessandroMiola]

I'd suggest to try to have a look at #124.
Hope it may help!

[VijayV28]

Thank you very much.