Remove Symmetric Zero Point in Compressed Outputs #59
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
dbogunowicz
approved these changes
May 23, 2024
dbogunowicz
approved these changes
May 23, 2024
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Outdated suggestions cannot be applied.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.
Requested by @dsikka, including the zero point in the state dict when the quantization is symmetric complicates the loading process in vLLM. On the compressed-tensors side, we now only save the zero point to the state_dict if it contains non-zeros (not symmetric). On reload in SparseML, the zero_point tensor will be recreated during decompression
Testing
Updated unit tests for the compressors to ensure the zero_point is removed from the state_dict in symmetric mode. The reload unit test already confirms we correctly recreate the zero_point during decompression