Added conv + hard_sigmoid oneDNN fuse pass #36869
Merged
Conversation
This file contains 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
|
Thanks for your contribution! |
|
@piotrekobiIntel Please, review this PR |
lidanqing-intel
changed the title
ghost
approved these changes
Nov 2, 2021
|
@jczaja please merge |
ghost
pushed a commit
to piotrekobi/Paddle
that referenced
this pull request
Nov 3, 2021
…e#36869) * added conv + hard_sigmoid fuse pass * Removed IsOptional() statements * Reverted removing optional
jakpiase
changed the title
Sign up for free
to 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Outdated suggestions cannot be applied.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.
PR types
New features
PR changes
OPs
Describe
Added conv + hard_sigmoid oneDNN fuse pass. It is designed mainly for PPLCNets architectures. This fuse pass improves PPLCNet_x0_25 model performance by 3% and PPLCNet_x1_0 performance by 1.4%
OneDNN does not have pure hard_sigmoid activation, that's why 2 eltwise postops are needed.
hard_sigmoid(x, slope, offset) = max(0, min(1, slope * x + offset))With oneDNN, we can get the inner equation by
eltwise_linear(x, slope, offset) = slope * x + offsetAnd adding clip after that would give us the exact equation as pure hard_sigmoid
eltwise_clip(eltwise_linear(x, slope, offset), 0, 1) = max(0, min(1, slope * x + offset))