re-submit softmax_with_cross_entropy hard label (#35283) #35660
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
|
Thanks for your contribution! |
JiaXiao243
approved these changes
Sep 11, 2021
JiaXiao243
left a comment
JiaXiao243
left a comment
There was a problem hiding this comment.
LGTM. ResNet50 top1 acc result is 76.3%.
AnnaTrainingG
pushed a commit
to AnnaTrainingG/Paddle
that referenced
this pull request
Sep 29, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.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.
PR types
Performance optimization
PR changes
Ops
Describe
Op softmax_with_cross_entropy optimization hard label. This PR includes:
hard label forward kernel: use same idea of softmax's implementation: SoftmaxWithCrossEntropyHardLabel
hard label backward kernel: merge multiple kernels to one kernel: SoftmaxWithCrossEntropyGradHardLabel
This PR re-submit PR 32290, which was reverted by PR 33340. The difference is from manual written kernel and cudnn implementation (log softmax)
This PR also modifies the absolute error check of "parallel_margin_cross_entropy". The owner (Guoxia Wang) is approved.