Add a doc on how to update PyTorch version

[huydhn]

Purpose

Add a doc on how to update PyTorch version for future PyTorch releases using #16859 as an example.

This is to address questions coming from #19705 where we plan to test 2.8 with vLLM before its release

cc @zou3519 @atalman

[github-actions]

👋 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 fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

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 ready label to the PR or enable auto-merge.

🚀

[gemini-code-assist]

Summary of Changes

Hello @huydhn, 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 crucial new documentation page that standardizes the complex process of updating PyTorch versions within the vLLM project's continuous integration and deployment pipelines. The guide aims to streamline future PyTorch upgrades by providing clear, actionable steps and solutions for common challenges, ensuring vLLM remains compatible with the latest PyTorch releases.

Highlights

• New Documentation: A comprehensive guide docs/ci/update_pytorch_version.md has been added to detail the process of updating PyTorch versions within vLLM's CI/CD.
• PyTorch Version Update Process: The document outlines best practices for testing PyTorch Release Candidates (RCs), managing different CUDA versions, and addressing common challenges like long build times and updating dependent libraries (e.g., FlashInfer, xFormers) by building them from source.
• CI/CD Workflow Improvements: It provides practical steps and workarounds for ensuring smooth PyTorch upgrades, including using --extra-index-url for specific CUDA versions and a temporary solution for sccache warm-up to mitigate build timeouts.

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

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 adds a comprehensive guide on updating PyTorch versions in vLLM. The review comments focus on minor typographical errors, ensuring consistency in tool usage, clarifying potentially ambiguous information, and enhancing document usability through linked references. These suggestions aim to improve the document's clarity, accuracy, and long-term maintainability.

[mgoin]

Nice write up!

[houseroad]

Thanks a lot for the detailed doc.