[BugFix] Fix DeepGEMM warmup, no m.weight_scale_inv #25206
Merged
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
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request provides a critical bug fix for the DeepGEMM warmup process. The change correctly updates the attribute access from m.weight_scale_inv to m.weight_scale in _extract_data_from_linear_base_module. This is necessary because Fp8LinearMethod deletes the weight_scale_inv attribute and creates weight_scale after processing the weights. Without this fix, an AttributeError would occur during model warmup for models using block-wise FP8 quantization, preventing them from being used. The fix is correct and essential.
tlrmchlsmth
approved these changes
Sep 18, 2025
ywang96
pushed a commit
to ywang96/vllm
that referenced
this pull request
Sep 19, 2025
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
debroy-rh
pushed a commit
to debroy-rh/vllm
that referenced
this pull request
Sep 19, 2025
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
chenxi-yang
pushed a commit
to chenxi-yang/vllm
that referenced
this pull request
Sep 20, 2025
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com> Signed-off-by: Chenxi Yang <cxyang@fb.com>
FeiDaLI
pushed a commit
to FeiDaLI/vllm
that referenced
this pull request
Sep 25, 2025
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
charlifu
pushed a commit
to ROCm/vllm
that referenced
this pull request
Sep 25, 2025
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com> Signed-off-by: charlifu <charlifu@amd.com>
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.
#21404 broke DeepGEMM warm-up, small fix