[misc] update swift patch_conv3d

[Jintao-Huang]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @Jintao-Huang, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request focuses on enhancing the robustness and accuracy of the inference engine by improving how sequence lengths and attention masks are handled during embedding and reranker score extraction. It includes updates to example scripts for API consistency, adds a PyTorch version compatibility safeguard for a Conv3d patch, and provides minor documentation clarifications to assist users with environment variable configurations.

Highlights

• Improved Inference Logic: The pull request refines the logic for extracting embeddings and reranker scores by consistently utilizing attention_mask to identify the last valid token in sequences, leading to more accurate results, especially for padded inputs.
• API Parameter Alignment: Example scripts for embedding and reranker tasks have been updated to use attn_impl instead of attn_implementation, aligning with a likely change in the underlying API.
• PyTorch Version Compatibility: A compatibility check has been added to the _patch_conv3d function, ensuring that the custom Conv3d forward method behaves correctly across different PyTorch versions, specifically providing backward compatibility for versions older than 2.9.0.
• Documentation Updates: The documentation for command-line parameters has been updated to clarify the compatibility of qwen3_vl environment variables with those used by qwen2_5_vl, including automatic conversion for variables like MAX_PIXELS and VIDEO_MAX_PIXELS.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces several improvements. It adds a compatibility fix for patch_conv3d to support newer PyTorch versions, which is a crucial update. The refactoring to use the get_last_valid_indices utility function in pt_engine.py and patcher.py is a great change that improves code clarity and maintainability by centralizing the logic for handling padded sequences. The parameter rename from attn_implementation to attn_impl is applied consistently. Overall, these are solid changes. I have one minor suggestion to improve code formatting in an example file for better readability.

[gemini-code-assist]

For better readability and adherence to PEP 8 style guidelines (which recommend a line length of 79-99 characters), it's better to format the arguments to PtEngine across multiple lines, as it was before this change.

        'Qwen/Qwen3-Embedding-4B',
        task_type='embedding',
        torch_dtype=torch.float16,
        attn_impl='flash_attention_2')