Skip to content

[bugfix] fix qwen3_omni infer#9164

Merged
Jintao-Huang merged 3 commits into
modelscope:mainfrom
Jintao-Huang:fix_qwen3_omni_infer
Apr 21, 2026
Merged

[bugfix] fix qwen3_omni infer#9164
Jintao-Huang merged 3 commits into
modelscope:mainfrom
Jintao-Huang:fix_qwen3_omni_infer

Conversation

@Jintao-Huang
Copy link
Copy Markdown
Collaborator

No description provided.

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request modifies the _get_position_ids method in the Qwen2VLTemplate class to return an empty dictionary during inference. However, the review points out that this change does not affect qwen3_omni as intended because that model overrides the method elsewhere, and the new logic makes subsequent code redundant.

Comment thread swift/template/templates/qwen.py Outdated
@Jintao-Huang
Copy link
Copy Markdown
Collaborator Author

/gemini review

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates the documentation by adding a link to the benchmark shell script in both the Chinese and English versions of the Quick-start guide. It also modifies the _get_position_ids method in the Qwen template to return an empty dictionary when not in training mode. I have no feedback to provide as there were no review comments to evaluate.

@Jintao-Huang Jintao-Huang merged commit 2b1588f into modelscope:main Apr 21, 2026
2 of 3 checks passed
Jintao-Huang added a commit that referenced this pull request Apr 23, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ms-swift 4.2.0 dev + transformer 5.5.0 在Qwen3-Omni-30B-A3B-Instruct中推理文本+视频报错

2 participants