fix: honor weight-check skips for quantized entries - #32809
Merged
Conversation
Propagate skip_compare_names through the quantized CheckEntry path so skip_tensor_list excludes quantized weights from comparison and checksums. Restore the matching CUDA regression and update the checksum E2E assertion for the per-role parallelism_info wire shape.
Contributor
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
Collaborator
Author
This was referenced Jul 29, 2026
Zhichenzzz
approved these changes
Jul 29, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Restore weight-check skip semantics and per-role checksum assertions after the v0.5.16 rebase.
Symptom & Reproduction
skip_tensor_listwas still compared.test_skip_set_marks_quantized_entry_not_comparedobservedshould_compare=Truebefore the fix.Root Cause
_build_check_entrieshard-codedTruefor quantized entries.test_weight_checker_e2e.pyretained the pre-listparallelism_infoassertion.Fix
Use
name not in skip_compare_namesfor quantized entries and restore the omitted CUDA and E2E assertions.Verification
test_skip_set_marks_quantized_entry_not_comparedverifies a skipped quantized entry hasshould_compare=False.test_weight_checker.pypasses all 60 tests.ruffpasses.blackandisortpass.git diff --checkpasses.sglang-kernel==0.4.4is below required0.4.5.Review Focus
_build_check_entriesto confirm the raw non-persistent-buffer behavior remains unchanged.test_e_checksum_returns_ranks_with_hashesagainst the currentList[ParallelismInfo]wire shape.CI States
Latest PR Test (Base): ❌ Run #30487047856
Latest PR Test (Extra): ❌ Run #30487047560