Open starhiking opened 4 years ago
It really works, but why?
It really works, but why?
CUDA is avaliable for that situation
It really works, but why?
CUDA is avaliable for that situation
thank you, I am not familiar with that, but I understand it now.
Was a PR created for this ? This really fixes an error in the code.
The original code doesn't consider the gpu situation for loss, and will lead to the problem:
RuntimeError: Expected object of backend CUDA but got backend CPU for ...
original code: https://github.com/yhenon/pytorch-retinanet/blob/ef2c16f4ad66b652b836be36465ee63c978f3194/retinanet/losses.py#L64
change:
regression_losses.append(torch.tensor(0).float().cuda())