[Tensorflow] Fix conv2d_transpose for NHWC layout#14546
[Tensorflow] Fix conv2d_transpose for NHWC layout#14546echuraev merged 7 commits intoapache:mainfrom jikechao:main
Conversation
If "data_format" == "NHWC", the kernel_layout should be "HWOI" rather than "HWIO".
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.
Generated by tvm-bot |
|
As we can see in frontend about conv2d_transpose @leandron @vvchernov Please help me review this PR. Thanks! |
|
@jikechao it would be good to add test cases for this fix |
|
@yongwww @vvchernov Thanks for your review. Thus, the condition |
|
@jikechao Thank you for your PR.
The link in your previous message is broken and doesn't highlight any line in the code. Also, could you please add a unit test for your changes? |
|
Hello @jikechao! Thank you! You are right about condition fix, but it is still needed CI test to check that it works ok now and prevent further issue around it. |
|
@echuraev Indeed, this link can't redirect to the highlight place by clicking it directly. As a workaround, you can open this link by copying this link and past it into a new browser window. I'll add a unit test sooner! |
echuraev
left a comment
There was a problem hiding this comment.
LGTM. Thanks. Please fix CI and then it can be merged.
|
@echuraev CI fixed, could you help me merge this PR? |
|
@jikechao Thank you for your work :) |
If "data_format" == "NHWC", the kernel_layout for conv2d_transpose should be "HWOI" rather than "HWIO".