Don't use RawReducer for activation shape collection in Fast Bias Correction #3642
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
daniil-lyakhov
approved these changes
Sep 2, 2025
Co-authored-by: Daniil Lyakhov <[email protected]>
Merged
3 tasks
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Outdated suggestions cannot be applied.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.
Changes
As in the title.
Reason for changes
This PR reduces memory footprint when applying Fast Bias Correction algorithm: collecting raw activations is not required to obtain their shapes. Avoiding using raw reducers allows to save some memory otherwise allocated for the activations.
Example quantization run on vision encoder from
OpenGVLab/InternVL2-1Bwith 4 calibration data samples:Since there is no need to allocate so much memory, statistics collection time also improves.
Related tickets
172800
Tests
Existing tests cover the new changes.
https://ci-adas-icv.iotg.sclab.intel.com/view/all/job/NNCF/job/manual/job/post_training_quantization/714/artifact/results.html