Disable gradient inplace and its check #693
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
Format. Disable gradinet check Disable gradient inplace and its check Disable gradient inplace and its check Format
Write relu and add2 backward manually
KazukiYoshiyama-sony
added
the
release-note-op-layer
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Outdated suggestions cannot be applied.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.
Disabling the gradient inplace and its check saves more memory. In the backward pass, we can usually re-use the memory region the memory manage already cached, so this change normally does not increase the total memory usage in training. Because of the no inplace check for the gradient, we can utilize the inplace-operation in the forward pass, resulting in more efficiency in memory. We verified by the regression tests in various classification/regression networks and also the utilization of the inplace-operation in the forward pass without checking the inplace in gradients in memory usage.