Document Qwen3.5 FLA patch for NPU support#9237
Merged
Merged
Conversation
Added detailed explanation of the Qwen3.5 FLA patch for NPU support, including its functionality and impact on the transformers library.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request updates the documentation to include a detailed explanation of the Qwen3.5 linear attention patch for NPU support, outlining how it redirects Triton operators to MindSpeed implementations. A review comment identified a version mismatch for the 'flash-linear-attention' package in the documentation, which has been addressed with a correction.
| - 该 patch 主要覆盖的是 **Qwen3.5 linear attention 的 gated-delta-rule 路径**； | ||
| - 它并不等价于“将整个 fla 包完整替换为 MindSpeed”； | ||
| - 若需要这条路径生效，请确保当前环境中可以正确导入 MindSpeed。 | ||
| - 精度对齐验证版本：torch 2.7.1 + MindSpeed 0.12.1 + flash-linear-attention 4.1.0 + triton-ascend 3.2.0 + transformers 5.2.0 |
Contributor
There was a problem hiding this comment.
The version for flash-linear-attention is listed as 4.1.0, but the pull request description specifies 0.4.1. This discrepancy should be corrected to ensure accuracy for users setting up the environment.
Suggested change
| - 精度对齐验证版本：torch 2.7.1 + MindSpeed 0.12.1 + flash-linear-attention 4.1.0 + triton-ascend 3.2.0 + transformers 5.2.0 | |
| - 精度对齐验证版本：torch 2.7.1 + MindSpeed 0.12.1 + flash-linear-attention 0.4.1 + triton-ascend 3.2.0 + transformers 5.2.0 |
Collaborator
|
Please also supplement the English document, thanks. |
Jintao-Huang
approved these changes
Apr 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
3 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.
Added detailed explanation of the Qwen3.5 FLA patch for NPU support, including its functionality and impact on the transformers library.
PR type
PR information
verified version -- transformers==5.2.0 triton-ascend==3.2.0 flash-linear-attention==0.4.1 torch==2.7.1