[Build] Add upper limit for transformers version #20225
Conversation
This change adds an upper limit for the `transformers` package version because Whisper fails to load with the latest version (4.53.0). I'm not sure if the problem affects other models yet. Issue vllm-project#20224 Signed-off-by: Russell Bryant <[email protected]>
|
👋 Hi! Thank you for contributing to the vLLM project. 💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels. Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging. To run CI, PR reviewers can either: Add 🚀 |
![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 @russellb, 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 introduces a critical dependency version constraint to resolve a known issue where the transformers library's latest version causes failures, particularly when loading the Whisper model. By pinning the maximum allowed version, it ensures the stability and functionality of the build system and dependent models.
Highlights
- Dependency Management: An upper version limit has been added for the
transformersPython package inrequirements/common.txt. - Issue Mitigation: This change specifically addresses an issue where the Whisper model fails to load with
transformersversion 4.53.0, as detailed in issue #20224, by preventing the use of this problematic version.
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 aims to add an upper version limit for the transformers package to prevent an issue with Whisper models. My review found a potential critical issue where the implemented version constraint seems to contradict the problem description, potentially making the fix ineffective. I've left a comment with a suggestion to address this.
| blake3 | ||
| py-cpuinfo | ||
| transformers >= 4.51.1 | ||
| transformers >= 4.51.1, <= 4.53.0 # Upper limit because of issue #20224 |
There was a problem hiding this comment.
There seems to be a discrepancy between the pull request description and the implemented change.
The description states:
Whisper fails to load with the latest version (4.53.0).
However, the version constraint added is ... <= 4.53.0, which allows pip to install version 4.53.0. If this version is indeed faulty, this change will not resolve the issue.
To exclude the problematic version, the constraint should likely be < 4.53.0.
Could you please confirm if version 4.53.0 is the one causing the failure? If so, please update the version specifier.
transformers >= 4.51.1, < 4.53.0 # Upper limit because of issue #20224
|
Resolved by 20244
needed to be |
This change adds an upper limit for the
transformerspackage versionbecause Whisper fails to load with the latest version (4.53.0). I'm not
sure if the problem affects other models yet.
Issue #20224
Signed-off-by: Russell Bryant [email protected]