-
Notifications
You must be signed in to change notification settings - Fork 1.4k
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.
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
New style convolution_2d and deconvolution_2d #3163
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please check travis result.
else: | ||
return func(x, W, b) | ||
func = Deconvolution2DFunction(stride, pad, outsize) | ||
args = x, W |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How about following code?
if b is None:
args = x. W
else:
args = x, W, b
Please rebase or merge master |
LGTM except comment. |
184423a
to
c75cd7b
Compare
Rebased. |
**kwargs): | ||
cover_all = None | ||
conv_desc = None | ||
filter_desc = None |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do you cache this descriptor?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I didn't change the behavior of caching descriptors. Is it better to remove it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Note: the above change is just about the default values for these attributes.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think deleting it is okay.
Removed the descriptors. |
LGTM! |
This PR implements the new style version of
F.convolution_2d
andF.deconvolution_2d
. These functions are now twice differentiable (but not three times differentiable until some functions used in backward, e.g.F.sum
, support double backward).This PR depends on #3096, and is a part of #3147.