[New model] Qwen3-next support #2917
Open
+4,120
−170
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
|
👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:
If CI fails, you can run linting and testing checks locally according Contributing and Testing. |
![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 adds support for the new Qwen3-next model, which is a hybrid attention model. The changes are extensive, touching attention mechanisms, model runner logic, and adding new custom operators. While the implementation is comprehensive, I've identified several critical issues related to code duplication, performance bottlenecks in the new model's prefill implementation, and potential correctness issues due to hardcoded values. There are also some high-severity issues regarding dead code and missed performance optimizations. Addressing these points will significantly improve the robustness and performance of the new model support.
wangxiyuan
force-pushed
the
qwen3-next
branch
from
271794a to
73a8e05
Compare
Signed-off-by: MengqingCao <cmq0113@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: MengqingCao <cmq0113@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: Angazenn <supperccell@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: Angazenn <supperccell@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: Angazenn <supperccell@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: Angazenn <supperccell@163.com> Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
force-pushed
the
qwen3-next
branch
from
626d6b8 to
c78e773
Compare
wangxiyuan
added
ready
Angazenn
reviewed
Sep 14, 2025
| # print(f"self.layer_idx: {self.layer_idx}, 111 mixed_qkv_non_spec: {mixed_qkv_non_spec}") | ||
|
|
||
| # 2.1: process the mutli-query part | ||
| # if spec_sequence_masks is not None: |
There was a problem hiding this comment.
spec_sequence_masks is not None is used to choose non-MTP branch. Since MTP for qwen3 next is not supported on npu now, maybe we can check it here.
Yikun
reviewed
Sep 14, 2025
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Yikun
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
force-pushed
the
qwen3-next
branch
from
b46d843 to
b4a3566
Compare
wangxiyuan
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
force-pushed
the
qwen3-next
branch
from
35e027f to
706348d
Compare
wangxiyuan
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
added
ready-for-test
wangxiyuan
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
added
ready-for-test
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
wangxiyuan
force-pushed
the
qwen3-next
branch
from
3f6e838 to
81bb245
Compare
wangxiyuan
added
ready-for-test
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.
What this PR does / why we need it?
Add Qwen3-next support.
Does this PR introduce any user-facing change?
Yes, users can use Qwen3 next.
Related doc: #2916 the tutorial will be ready in here
How was this patch tested?
Doc CI passed
Related: #2884