Skip to content

identity hop shape fix #59

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
matteopresutto wants to merge 3 commits into
base: master
Choose a base branch
from
Open

identity hop shape fix #59

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
41ed700
identity hop shape fix
matteopresutto Apr 23, 2016
adbe865
Update Deep_Residual_Learning_CIFAR-10.py
matteopresutto Jul 5, 2016
f0198fa
Update Deep_Residual_Learning_CIFAR-10.py
matteopresutto Jul 5, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion papers/deep_residual_learning/Deep_Residual_Learning_CIFAR-10.py
100644 → 100755
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,7 +113,7 @@ def residual_block(l, increase_dim=False, projection=False):
block = NonlinearityLayer(ElemwiseSumLayer([stack_2, projection]),nonlinearity=rectify)
else:
# identity shortcut, as option A in paper
identity = ExpressionLayer(l, lambda X: X[:, :, ::2, ::2], lambda s: (s[0], s[1], s[2]//2, s[3]//2))
identity = ExpressionLayer(l, lambda X: X[:, :, ::2, ::2], lambda s: (s[0], s[1], int((s[2] + 0.5) // 2), int((s[3] + 0.5) // 2)))
Copy link
Member

@f0k f0k Jul 5, 2016 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wait, it should have been (s[2]+1)//2. This rounds up and doesn't need an int conversion either. Adding +.5 before the division was wrong, it should have been after the division as in int(s[2]/2. + .5), sorry.

padding = PadLayer(identity, [out_num_filters//4,0,0], batch_ndim=1)
block = NonlinearityLayer(ElemwiseSumLayer([stack_2, padding]),nonlinearity=rectify)
else:
Expand Down