Repository navigation
Conversation
vllm-project#59685 made DSparkDeepseekV4ForCausalLM.load_weights read self.model.layers[0].ffn.use_mega_moe, which the test's stub model lacks. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: khluu <khluu000@gmail.com>
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
mergify
Bot
added
deepseek
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Overview
Fixes
:amd: (MI355) Basic Models (Other)on main. It is non-optional and has been red since #59685 ([ROCm][DSv4] AITER MegaMoEV2) merged. Example: https://buildkite.com/vllm/ci/builds/93397#01a11807-77e4-4894-94e2-b18903ce011cClaims
tests/models/test_deepseek_v4_dspark_rocm.pypasses again: 6/6, up from 4/6 on main.Validation
I ran the file in the CI image for main commit
08567505b0(vllm-ci-postmerge-repo:08567505b0…) on one H200. The file has a module-level import that needs a visible GPU; the tests themselves arecpu_teststubs.2 failed, 4 passed:AttributeError: 'types.SimpleNamespace' object has no attribute 'layers'. This is the same error as the MI355 job.6 passedCommand:
pytest -q models/test_deepseek_v4_dspark_rocm.py. I did not run it on MI355; the PR's MI355 job will confirm.Details
#59685 changed
DSparkDeepseekV4ForCausalLM.load_weights(vllm/models/deepseek_v4/amd/dspark.py:395) to choose the routed-experts prefix fromself.model.layers[0].ffn.use_mega_moe. The twoload_weightstests build aSimpleNamespacestub formodel.modelthat has onlyconfidence_head, so the new attribute access fails. The stub now also has one layer withuse_mega_moe=False, which matches the non-MegaMoE default. The tests monkeypatchfused_moe_make_expert_params_mapping, so the prefix value doesn't affect what they assert.#59685's own CI didn't catch this because its changes didn't trigger the MI355 Basic Models (Other) job. #59050 added this test file to that job.
Not a duplicate: I searched open PRs for
dspark_rocm/test_deepseek_v4_dspark_rocmand found none.Pull Request Checklist
I used vLLM's
/pr-checklistskill. (Mandatory for agents, optional for humans).AI assistance was used during the creation of this PR. Claude Code wrote it; @khluu has not reviewed it yet.
Design Fit: Minimizes impact on core components, reuses existing functionality, and justifies added complexity.
Testing and Validation: Validates the change and ensures any added tests are meaningful and reliable, with CI coverage or documented CI resource constraints and validation performed outside CI.
Code Quality and Style: Keeps code and comments clear and concise, and updates relevant documentation and examples.
Pull Request Contents: Includes a brief summary and relevant links, supports claims with evidence, explains root causes and implementation trade-offs, and follows the contributing guide.
🤖 Generated with Claude Code