I heard about Depthwise conv on pytorch that is slower than common conv because of optimization issue.
And I found a user implementation of depthwise conv based on pytorch. The link is below:
My question is "Is it okay to use this implementation on my model if i want my model to get officially ranked?"
Thank you.Posted by: GWJ @ July 2, 2020, 4:17 a.m.
It is OK.Posted by: cszn @ July 2, 2020, 11:36 a.m.