Uh oh!
There was an error while loading. Please reload this page.
Fix group transpose conv2d - #9443
Closed
alicja-SiMa-ai wants to merge 1 commit into
Closed
Conversation
alicja-SiMa-ai
requested review from
Huyuwei, Laurawly, MarisaKirisame, ZihengJiang, anijain2305, areusch, comaniac, icemelon, jcf94, jroesch, junrushao, jwfromm, kevinthesun, masahi, mbrookhart, merrymercy, slyubomirsky, tqchen, vinx13, wweic, yzhliu and zhiics
as code ownersNovember 4, 2021 14:25
Change includes topi implementation, tests, generic and x86 strategy for group transposed convolution. Signed-off-by: Alicja Kwasniewska <alicja.kwasniewska@sima.ai>
alicja-SiMa-aiforce-pushed
the
fix/group_transpose_conv2d
branch
from
November 5, 2021 04:18
7aa2efa to
8eb15b7CompareAndrewZhaoLuo
commented
Nov 5, 2021
Contributor
Hmm is this perhaps related to #9336 ? |
alicja-SiMa-ai
commented
Nov 5, 2021
ContributorAuthor
@AndrewZhaoLuo#9336 changes the layout conversion. This PR addresses a different issue, it adds implementation of group transpose convolution 2D which was missing and fixes weight shape to support groups. |
vinx13
reviewed
Nov 8, 2021
vinx13
left a comment
Member
There was a problem hiding this comment.
would be great to also add Relay tests here https://github.com/apache/tvm/blob/main/tests/python/relay/test_op_level2.py#L818
junrushao
commented
Nov 12, 2021
Member
@Lyken17@alicja-SiMa-ai just to double check, is this commit superseded via this commit 3ad7c4a? Thanks! |
alicja-SiMa-ai
commented
Nov 12, 2021
ContributorAuthor
Yes @junrushao1994, it was integrated into #9465. Closing this one. Thanks everyone! |
junrushao
commented
Nov 12, 2021
Member
Thank you @alicja-SiMa-ai for the huge effort!! This is super helpful! |
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
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Shape of weights was incorrectly set for grouped transposed conv2D and implementation of the grouped transposed conv2D was missing. As a result, some models were not supported. Please see example of the CenterNet model here with error message
in particular dimension 1 conflicts 0 does not match 1; unable to unify: Tensor[(256, 0, 4, 4), float32] and Tensor[(256, 1, 4, 4), float32].This PR adds missing implementation and required tests.
cc @comaniac