Fix broken conv/deconv reshaping #3096
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
LayerSetUp This also eliminates the extra copying of bottom's shape.
9e6d33e to
6a00eca
Compare
Member
|
This looks right to me but @jeffdonahue should check as the original author. |
Contributor
|
Looks reasonable to me, but hard to know without any tests. Since this is a bugfix though, I think this can be merged given the existing tests pass and that you've successfully used this for reshaping in your work @longjon. |
Contributor
Author
|
I'll go ahead and merge then since this passes the tests and a cursory check which the old code failed, but we should still fix up the tests so that this is checked more carefully. |
longjon
added a commit
that referenced
this pull request
Sep 21, 2015
Fix broken conv/deconv reshaping
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
#2049 moved reading and storing of
bottomshapes toLayerSetUp, which (silently) breaks reshaping to different input shapes.This PR fixes that issue. I'm also not sure why another blob was introduced to store the input shape; this PR uses an inline function that wraps
bottom[0]->shapeinstead. (Note that that's now somewhat awkwardly stored asbottom_shape_in order to get around the fact thatbottomis neither a member variable nor argument to functions to such ascompute_output_shape.)Again, this should definitely be tested, but this PR is just the fix...