Conversation
maxjeblick
reviewed
Mar 19, 2025
Signed-off-by: SimJeg <[email protected]>
Signed-off-by: SimJeg <[email protected]>
SimJeg
force-pushed
the
simon/duo-attention-on-the-fly
branch
from
8152e87 to
c844335
Compare
Signed-off-by: SimJeg <[email protected]>
Signed-off-by: SimJeg <[email protected]>
Signed-off-by: SimJeg <[email protected]>
Collaborator
Author
|
Here are the results: not only the new method is faster (from "several hours on an 8 GPU node" to a few seconds on 1 GPU), but it's also better ! The on-the-fly technique requires 30sec with 50 samples but I obtained similar results using only 10 samples.
|
Collaborator
Author
|
Additional results with 5 samples instead of 50 to compute head scores on the fly: |
Signed-off-by: SimJeg <[email protected]>
Signed-off-by: SimJeg <[email protected]>
maxjeblick
approved these changes
Mar 19, 2025
giulio98
pushed a commit
to miriam-16/kvpress
that referenced
this pull request
Apr 4, 2025
Signed-off-by: giulio98 <[email protected]>
maxjeblick
pushed a commit
that referenced
this pull request
Aug 12, 2025
Signed-off-by: Max Jeblick <[email protected]>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR fixes #62 and propose a new way to compute DuoAttention scores for classification of streaming and retrieval heads. I will report results based on this PR in this branch.