fix bytes2tensor error in torch version == 2.6.0. #950
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 @hiworldwzj, 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 resolves a specific compatibility issue with PyTorch version 2.6.0 by adjusting the parameters used when loading tensors from byte streams. The modification ensures that the bytes2tensor utility functions correctly, preventing errors related to torch.load's behavior in the specified PyTorch version.
Highlights
- PyTorch Compatibility Fix: Updated the
bytes2tensorutility function inlightllm/server/embed_cache/utils.pyto explicitly passweights_only=Falsetotorch.load. This change addresses an error encountered when using PyTorch version 2.6.0, ensuring proper deserialization of tensors from byte streams.
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 is currently in preview and 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 to provide feedback.
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 forward-compatibility issue with an upcoming PyTorch release by explicitly setting weights_only=False in a torch.load call. The change is correct and necessary to prevent the application from breaking with newer versions of PyTorch. My review includes a suggestion to add a comment explaining the security implications of this setting and clarifying why it is safe in this particular context, which will improve long-term maintainability and code clarity.
| @@ -22,7 +22,7 @@ def tensor2bytes(t: torch.Tensor): | |||
|
|
|||
| def bytes2tensor(b): | |||
| # return torch.from_numpy(np.frombuffer(b, dtype=np.float16)).cuda() | |||
| return torch.load(BytesIO(b)) | |||
| return torch.load(BytesIO(b), weights_only=False) | |||
There was a problem hiding this comment.
Explicitly setting weights_only=False is the correct fix for the breaking change in upcoming PyTorch versions where this argument will default to True. However, this setting can introduce a security vulnerability if the serialized data comes from an untrusted source, as it can lead to arbitrary code execution during deserialization.
In this specific case, the data is generated internally by the tensor2bytes function and passed via shared memory, so it can be considered trusted. To improve code clarity and prevent future misuse by developers who may not be aware of the context, I recommend adding a comment explaining why weights_only=False is necessary and why it's considered safe here.
| return torch.load(BytesIO(b), weights_only=False) | |
| return torch.load(BytesIO(b), weights_only=False) # Required for unpickling tensor objects; safe as data is generated internally. |
No description provided.