Explicitly set batch=1 for NNCF in order to avoid issue with Wave2Vec #312
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
|
@AlexKoff88 @yujiepan-work @vuiseng9 please take a look |
ljaljushkin
commented
May 9, 2023
AlexKoff88
reviewed
May 10, 2023
AlexKoff88
approved these changes
May 10, 2023
|
@echarlaix Could this be merged before the upcoming dot release? |
|
The documentation is not available anymore as the PR was closed or merged. |
|
echarlaix
approved these changes
Jun 7, 2023
There was a problem hiding this comment.
Thanks for the fix @ljaljushkin ! Can you rebase as well as run the make style command to fix the tests
Done |
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Outdated suggestions cannot be applied.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.
What does this PR do?
Explicitly set batch size equal 1 in the NNCFConfig in order to avoid issue with Wave2Vec with the upcoming NNCF release.
With switching to NNCF v2.5.0, JPQD uses automatic structured pruning algorithm and currently it doesn't work properly with batch > 1 for Wave2Vec model.
It works fine when batch is 1. With this PR it is set in the NNCFConfig only, it should not affect training and inference with different batch sizes, since it's only used for NNCF internals to set up compression.
The proper fix for the Wave2Vec will be introduced in the next releases of NNCF. Here's the draft (openvinotoolkit/nncf#1784) that will be finalized later.
The fix is checked on the existing NNCF, JPQD tests.
Fixes # (issue)
Ticket 110126 in the internal NNCF tracking system