fix blockwise controlnet training by avoid inplace#844
fix blockwise controlnet training by avoid inplace#844Artiprocher merged 1 commit intomodelscope:mainfrom
Conversation
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary of Changes
Hello @mi804, 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 critical bug in the blockwise ControlNet training pipeline within the qwen_image.py file. The core issue stemmed from an unintended inplace modification of the 'image' tensor during the blockwise_forward operation, which could lead to training failures or incorrect behavior. The solution implemented ensures that a copy of the tensor slice is used for processing, thereby preventing direct modification of the original tensor and stabilizing the training process.
Highlights
- Blockwise ControlNet Training Fix: Resolved an issue where blockwise ControlNet training failed due to an inplace operation on the 'image' tensor. The fix involves creating a clone of the relevant image slice before processing it with 'blockwise_forward' to ensure operations are not performed in-place.
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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.
| 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 issue 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
-
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. ↩
There was a problem hiding this comment.
Code Review
This pull request addresses a bug in blockwise ControlNet training by avoiding an in-place operation on the image tensor. The previous implementation could lead to errors during backpropagation because a tensor was being modified in-place while also being used for gradient computation.
The fix correctly resolves this by first creating a clone() of the relevant image slice. This new tensor is then used in the blockwise_controlnet.blockwise_forward call and the subsequent addition. This change ensures that the gradient computation graph is not broken, allowing training to proceed correctly. The implementation is clear and effectively solves the issue.
fix blockwise controlnet training by avoid inplace
2 participants
No description provided.