Wall-clock performance issues
verbose-void opened this issue · 2 comments
verbose-void commented
LambdaResNet50 in comparison to a normal ResNet50 from my testing is 7x slower. I am training on ImageNet, and I am comparing your LambdaResNet50 to a torch.hub ResNet50.
Is this to be expected?
leaderj1001 commented
Perhaps the torch.einsum computation time is too slow.
I will find another way to reduce the computation time.
Thanks
verbose-void commented
yeah, that was the conclusion i came to as well. it's significantly faster than torch's einops still.