fix wan i2v/ti2v train bug

[huarzone]

fixes #1198, #1175

[gemini-code-assist]

Summary of Changes

Hello @huarzone, 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 addresses a training bug in the 'wan i2v/ti2v' pipeline by refining how the first frame of a video sequence is handled during the loss computation. The changes ensure that if a fixed first frame is provided, it is correctly incorporated into the latent inputs and subsequently excluded from the training target and noise prediction for loss calculation, which is crucial for stable and accurate video generation training.

Highlights

• Conditional First Frame Handling: Introduced conditional logic within the FlowMatchSFTLoss function to specifically handle first_frame_latents if they are provided in the input.
• Latent Input Modification: When first_frame_latents are present, the first frame of inputs["latents"] is explicitly set to these provided latents, ensuring a fixed or reference first frame.
• Loss Calculation Adjustment: The noise_pred and training_target are now sliced to exclude the first frame from the loss calculation when first_frame_latents are used, preventing the model from trying to predict or learn from this fixed frame.

🧠 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 a fix for a training bug in i2v/ti2v models. The change correctly handles conditioning on the first frame by modifying the input latents and adjusting the training target and prediction before loss calculation. The logic is sound and effectively resolves the issue. I have one minor suggestion to improve code clarity by avoiding a repeated conditional check.

[gemini-code-assist]

The condition "first_frame_latents" in inputs is checked twice in this function (here and on line 16). To improve readability and avoid this repetition, consider storing the result of this check in a boolean variable before the first if block and reusing that variable in both conditional statements.