Inverse of a tensor

1

Hi everyone, I am trying to do the inverse of a tensor element-wise.
like this:

sample[i].pow_(-1)

Unfortunately, the precision that I am looking for is not achieved although is float64 type.

Data example:
Having this tensor:
tensor([[[[1631., 1635., 1639., …, 4023., 4025., 4027.],
[1629., 1634., 1638., …, 4020., 4022., 4024.],
[1627., 1632., 1636., …, 4017., 4019., 4021.],
I get this result:
tensor([[[[0.0006, 0.0006, 0.0006, …, 0.0002, 0.0002, 0.0002],
[0.0006, 0.0006, 0.0006, …, 0.0002, 0.0002, 0.0002],
[0.0006, 0.0006, 0.0006, …, 0.0002, 0.0002, 0.0002]

I don’t know how to fix it… I thought that would be a cast conversion problem but I cannot find the problem.

2

Your output might just be cropped and you could increase the printed decimals via torch.set_printoptions(precision=10).