fix[example/hunyunvideo]:HunyuanVideo is compatible with MindSpore 2.6 and 2.7

[iugoood]

What does this PR do?

HunyuanVideo is compatible with MindSpore 2.6 and 2.7

1.example/hunyunvideo added GroupNorm class to fix the misalignment of GroupNorm precision types during model training and inference.
2. fix some bug

Fixes # (issue)

Adds # (feature)

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you make sure to update the documentation with your changes? E.g. record bug fixes or new features in What's New. Here are the
  documentation guidelines
• Did you build and run the code without any errors?
• Did you report the running environment (NPU type/MS version) and performance in the doc? (better record it for data loading, model inference, or training tasks)
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

@xxx

[gemini-code-assist]

Summary of Changes

Hello @iugoood, 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 primarily focuses on enhancing the HunyuanVideo project's compatibility with recent MindSpore versions (2.6 and 2.7). It includes critical bug fixes such as introducing a custom GroupNorm implementation to resolve precision issues, adjusting text encoder logic for accurate hidden state retrieval, and updating key dependencies. These changes ensure the model operates correctly and efficiently within the specified MindSpore environments.

Highlights

• MindSpore Compatibility: Updated HunyuanVideo to be compatible with MindSpore versions 2.6 and 2.7, along with corresponding CANN toolkit/kernel versions.
• GroupNorm Implementation: Introduced a custom GroupNorm class to address precision type misalignment during model training and inference, ensuring correct normalization behavior.
• Dependency Updates: Upgraded tokenizers to version 0.21.4 and transformers to version 4.50.0 to leverage newer features and bug fixes.
• Text Encoder Logic Fix: Corrected indexing logic for retrieving hidden states within the text encoder, specifically changing outputs[2] to outputs[1] in certain conditions.
• Training Script Argument Fix: Modified argument parsing in the training script (scripts/train.py) to correctly handle validation arguments by adding the -- prefix.

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 aims to make HunyuanVideo compatible with MindSpore 2.6 and 2.7. The changes include updating dependencies and environment configurations in the README and requirements.txt. It also introduces a custom GroupNorm class to handle data type precision issues and fixes some bugs related to argument parsing and activation function instantiation. My main concern is with the changes in hyvideo/text_encoder/__init__.py, where relying on tuple indexing to access model outputs is fragile and can lead to bugs depending on the model's configuration. I've left a comment with a suggestion for a more robust implementation.

[gemini-code-assist]

The type casting self.weight.to(x.dtype) and self.bias.to(x.dtype) is a good fix to ensure the affine parameters match the input tensor's data type, which resolves potential precision mismatches during training and inference.

[SamitHuang]

pls update the inference performance on ms2.6/2.7

[SamitHuang]

pls update the inference performance on ms2.6/2.7

@iugoood