Repository navigation
fix(container): backport vllm PR #40932 to unblock Qwen3-VL disagg multimodal #9522
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
saturley-hall
merged 1 commit into
release/1.2.0
from
qiwa/cherrypick-9491-backport-vllm-pr-40932
May 15, 2026
Merged
Changes from all commits
Commits
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
77 changes: 77 additions & 0 deletions
77
...m/patches/v0.20.1/0001-pr40932-remove-invalid-deepstack-boundary-check-for-qwen3-vl.patch
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,77 @@ | ||
| # SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. | ||
| # SPDX-FileCopyrightText: Copyright contributors to the vLLM project | ||
| # SPDX-License-Identifier: Apache-2.0 | ||
|
|
||
| From 22631f80a01a04b06398952e77d7890ab660ab10 Mon Sep 17 00:00:00 2001 | ||
| From: Isotr0py <mozf@mail2.sysu.edu.cn> | ||
| Date: Mon, 27 Apr 2026 15:27:06 +0800 | ||
| Subject: [PATCH] [Bugfix] Remove invalid deepstack boundary check for Qwen3-VL | ||
| (#40932) | ||
|
|
||
| Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn> | ||
| --- | ||
| vllm/model_executor/models/qwen3_omni_moe_thinker.py | 11 ----------- | ||
| vllm/model_executor/models/qwen3_vl.py | 11 ----------- | ||
| 2 files changed, 22 deletions(-) | ||
|
|
||
| diff --git a/vllm/model_executor/models/qwen3_omni_moe_thinker.py b/vllm/model_executor/models/qwen3_omni_moe_thinker.py | ||
| index ff8962c82..44e656ab8 100755 | ||
| --- a/vllm/model_executor/models/qwen3_omni_moe_thinker.py | ||
| +++ b/vllm/model_executor/models/qwen3_omni_moe_thinker.py | ||
| @@ -1778,11 +1778,6 @@ class Qwen3OmniMoeThinkerForConditionalGeneration( | ||
| return None # If vision tower is skipped | ||
| if getattr(self, "deepstack_input_embeds_num_tokens", 0) == 0: | ||
| return None | ||
| - if num_tokens > self.deepstack_input_embeds_num_tokens: | ||
| - raise ValueError( | ||
| - "Requested more deepstack tokens than available in buffer: " | ||
| - f"{num_tokens=} > {self.deepstack_input_embeds_num_tokens=}" | ||
| - ) | ||
|
|
||
| # get deepstack_input_embeds from buffer, and clear the buffer | ||
| return IntermediateTensors( | ||
| @@ -1824,12 +1819,6 @@ class Qwen3OmniMoeThinkerForConditionalGeneration( | ||
|
|
||
| # clear deepstack_input_embeds in buffer | ||
| if num_tokens > 0: | ||
| - if num_tokens > self.deepstack_input_embeds_num_tokens: | ||
| - raise ValueError( | ||
| - "Requested to clear more deepstack tokens than available in " | ||
| - "buffer: " | ||
| - f"{num_tokens=} > {self.deepstack_input_embeds_num_tokens=}" | ||
| - ) | ||
| for idx in range(self.deepstack_num_level): | ||
| self.deepstack_input_embeds[idx][:num_tokens].zero_() | ||
| self.deepstack_input_embeds_num_tokens = 0 | ||
| diff --git a/vllm/model_executor/models/qwen3_vl.py b/vllm/model_executor/models/qwen3_vl.py | ||
| index 4a0de15f0..2575f634b 100644 | ||
| --- a/vllm/model_executor/models/qwen3_vl.py | ||
| +++ b/vllm/model_executor/models/qwen3_vl.py | ||
| @@ -1716,11 +1716,6 @@ class Qwen3VLForConditionalGeneration( | ||
| return None # If vision tower is skipped | ||
| if getattr(self, "deepstack_input_embeds_num_tokens", 0) == 0: | ||
| return None | ||
| - if num_tokens > self.deepstack_input_embeds_num_tokens: | ||
| - raise ValueError( | ||
| - "Requested more deepstack tokens than available in buffer: " | ||
| - f"{num_tokens=} > {self.deepstack_input_embeds_num_tokens=}" | ||
| - ) | ||
|
|
||
| # get deepstack_input_embeds from buffer, and clear the buffer | ||
| return IntermediateTensors( | ||
| @@ -1762,12 +1757,6 @@ class Qwen3VLForConditionalGeneration( | ||
|
|
||
| # clear deepstack_input_embeds in buffer | ||
| if num_tokens > 0: | ||
| - if num_tokens > self.deepstack_input_embeds_num_tokens: | ||
| - raise ValueError( | ||
| - "Requested to clear more deepstack tokens than available in " | ||
| - "buffer: " | ||
| - f"{num_tokens=} > {self.deepstack_input_embeds_num_tokens=}" | ||
| - ) | ||
| for idx in range(self.deepstack_num_level): | ||
| self.deepstack_input_embeds[idx][:num_tokens].zero_() | ||
| self.deepstack_input_embeds_num_tokens = 0 | ||
| -- | ||
| 2.43.0 | ||
|
|
Oops, something went wrong.
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.
Uh oh!
There was an error while loading. Please reload this page.