Guillem Cucurull

Results 1 issues of Guillem Cucurull

When I use dice loss for Pytorch 1.3 I get the following error when I try to call `loss.backward()`. ``` RuntimeError: element 0 of tensors does not require grad and...