AttributeError: module 'catalyst.utils' has no attribute 'detach' while running Segmentation Tutorial

[faizankshaikh]

Describe the bug
When I run the tutorial on Segmentation, with little to no change from my side, I get the error AttributeError: module 'catalyst.utils' has no attribute 'detach' when I try to run the second cell in "Model Inference" of the notebook

To Reproduce
Run the notebook of Segmentation Tutorial on Colab and follow along with the steps

Expected behavior
The cell should show predictions from the model, i.e. image with the respective mask

Screenshots

[faizankshaikh]

Resolves if I add function before the cell is run and replace utils.detach with _detach

def _detach(tensor):
    return tensor.cpu().detach().numpy()

Can submit a PR by tomorrow if required. Found the solution hidden deep in the code

[Scitator]

Yeah sure, could you please add detach method to https://github.com/catalyst-team/catalyst/blob/master/catalyst/utils/torch.py ?

[faizankshaikh]

Done #766