[NPU] Add fused_linear_cross_entropy operator#1164
Open
lowdy1 wants to merge 2 commits intolinkedin:mainfrom
Open
[NPU] Add fused_linear_cross_entropy operator#1164lowdy1 wants to merge 2 commits intolinkedin:mainfrom
lowdy1 wants to merge 2 commits intolinkedin:mainfrom
Conversation
Contributor
Author
|
Tcc0403
reviewed
Mar 31, 2026
|
|
||
| # Here we calculate the gradient of logits_chunk in place so we can save memory. | ||
| # Grid size is capped at NPU core count; the kernel uses a grid-stride loop | ||
| liger_cross_entropy_kernel[(n_rows,)]( |
Collaborator
There was a problem hiding this comment.
Shouldn't we set grid size to num_cores?
Contributor
Author
There was a problem hiding this comment.
In liger_cross_entropy_kernel, there is no loop over num_programs() as seen in other kernels, so n_rows is used instead.
Collaborator
There was a problem hiding this comment.
Oh I thought we are using the ascend version?
Contributor
Author
There was a problem hiding this comment.
Sorry, my local branch was outdated. This has been addressed.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
To address the UB overflow issue observed in the benchmark, we introduced an operator with an NPU-friendly implementation of fused linear cross entropy. This fused operator relies on several underlying operations (e.g., large matrix multiplication, softmax, and cross entropy), so its current benchmark performance is not yet optimal. Further optimization may be needed.
Testing Done
Device: Atlas A2

python -m pytest ./test/transformers/test_fused_linear_cross_entropy.pymake testto ensure correctnessmake checkstyleto ensure code stylemake test-convergenceto ensure convergence