[FIX_FOR_VLLM_CUSTOM=4efd6ffde09477800294a8ed9cc752017812c3b1] Fix minimax_m2 import after mamba LINEAR refactor (+1 more)#1529
Closed
pawel-olejniczak wants to merge 2 commits into
Conversation
…nimax_m2 import after mamba LINEAR refactor Root cause: upstream vLLM #43556 relocated MiniMaxText01RMSNormTP out of vllm.model_executor.layers.mamba.linear_attn into the new vllm.model_executor.layers.minimax_rms_norm package. Upstream: vllm-project/vllm#43556 Fix: import MiniMaxText01RMSNormTP from minimax_rms_norm. Signed-off-by: Paweł Olejniczak <pawelx.olejniczak@intel.com>
…lti_model_api_server imports after serving-utils consolidation Root cause: upstream vLLM #44479 consolidated online serving utils, removing vllm/entrypoints/logger.py, vllm/entrypoints/openai/server_utils.py and vllm/entrypoints/utils.py. Upstream: vllm-project/vllm#44479 Fix: update three imports in multi_model_api_server.py to new canonical paths (RequestLogger -> serve.utils.request_logger; get_uvicorn_log_config -> serve.utils.server_utils; cli_env_setup/process_lora_modules -> serve.utils.api_utils). Signed-off-by: Paweł Olejniczak <pawelx.olejniczak@intel.com>
pawel-olejniczak
requested review from
PatrykWo,
adobrzyn,
afierka-intel,
iboiko-habana,
jbyczkow,
ksmusz,
mgawarkiewicz-intel,
michalkuligowski and
xuechendi
as code owners
pawel-olejniczak
requested a deployment
to
pre-merge-approval
GitHub Actions
Waiting
Contributor
There was a problem hiding this comment.
Pull request overview
Note
Copilot was unable to run its full agentic suite in this review.
Updates import paths to align with refactored module locations in vLLM/vLLM-Gaudi, ensuring these components resolve from their new packages.
Changes:
- Switched
MiniMaxText01RMSNormTPimport tovllm.model_executor.layers.minimax_rms_norm. - Updated OpenAI multi-model API server utility imports (
RequestLogger,get_uvicorn_log_config,cli_env_setup,process_lora_modules) tovllm.entrypoints.serve.utils.*.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| vllm_gaudi/models/minimax_m2.py | Adjusts RMSNorm import to the new MiniMax-specific module path. |
| vllm_gaudi/entrypoints/openai/multi_model_api_server.py | Updates server-related imports to new serve.utils locations. |
Collaborator
Author
|
Closing as duplicate. These two commits were consolidated into the existing rolling hourly-CI fix PR #1525 (per the single-rolling-PR rule). Only one open PR with the FIX_FOR_VLLM_CUSTOM prefix is allowed at a time. |
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.
This PR consolidates 2 hourly-CI fixes against vllm@
4efd6ffde09477800294a8ed9cc752017812c3b1per the single-rolling-PR rule (invariant I9).Bug 1: Fix minimax_m2 import after mamba LINEAR refactor
Root cause
Upstream vLLM moved
MiniMaxText01RMSNormTPout ofvllm.model_executor.layers.mamba.linear_attninto a dedicated modulevllm.model_executor.layers.minimax_rms_norm. The HPUminimax_m2model is eagerly imported byregister_model(), so the stale import broke every CI test at import time.Upstream PR
vllm-project/vllm#43556
Fix
Update the import of
MiniMaxText01RMSNormTPtovllm.model_executor.layers.minimax_rms_norm.Bug 2: Fix multi_model_api_server imports after serving-utils consolidation
Root cause
Upstream vLLM consolidated the online serving utilities, removing
entrypoints/logger.py,entrypoints/openai/server_utils.pyandentrypoints/utils.py. The HPU multi-model API server imported three symbols from those removed modules, breaking the unit-test import path.Upstream PR
vllm-project/vllm#44479
Fix
Repoint three imports in
multi_model_api_server.pyto the consolidated locations:serve.utils.request_logger(RequestLogger),serve.utils.server_utils(get_uvicorn_log_config) andserve.utils.api_utils(cli_env_setup, process_lora_modules).HPU verification
origin/main..HEAD) re-verified against vllm@4efd6ffde09477800294a8ed9cc752017812c3b1: PASSRelated PRs
None