Change bias initialization from 'embed' to 'heads'#371
Open
Conversation
csgoogle
force-pushed
the
fixbiassharding
branch
2 times, most recently
from
d822acb to
15af39f
Compare
entrpn
previously approved these changes
Apr 14, 2026
Collaborator
|
Could you add more details and results on the new commits? @csgoogle |
csgoogle
force-pushed
the
fixbiassharding
branch
5 times, most recently
from
9780b17 to
7a6ab88
Compare
csgoogle
force-pushed
the
fixbiassharding
branch
from
20430ab to
2ddf8ab
Compare
csgoogle
commented
Apr 15, 2026
| raise ValueError(f"Flash attention expects rank-3 or rank-4 inputs, got rank {tensor.ndim}.") | ||
|
|
||
|
|
||
| def _select_flash_block_sizes( |
Collaborator
Author
There was a problem hiding this comment.
just refactoring no logical changes and added unit tests, this would be helpful for other ulyesses attention pr
Perseus14
approved these changes
Apr 15, 2026
entrpn
approved these changes
Apr 15, 2026
Collaborator
Author
done |
|
🤖 Hi @csgoogle, I've received your request, and I'm working on it now! You can track my progress in the logs for more details. |
|
🤖 I'm sorry @csgoogle, but I was unable to process your request. Please see the logs for more details. |
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
doc: https://docs.google.com/document/d/1absFkpQAMM3YaYWxO_FYeqzDpypYeDbPsJRAV86nFQ0/edit?usp=sharing&resourcekey=0-FOzOmM0UdfU1LcDd_7epvw
Results
mainfixbiasshardingNotes
tp=1configs — improvement only surfaces when tensor parallelism is active, as the axis fixes reduce parameter all-gather overhead in MLP layerstp=4ortp=8where parameter communication is a larger fraction of step timeVideo Quality Comparison
mainfixbiasshardingPSNR/SSIM (frame-by-frame, 81 frames):
Video and Xprof after fix:
https://console.cloud.google.com/storage/browser/sagarchapara/shardingfixes