Increase vLLM health timeout to 10 minutes

[hamishivi]

Increase timeout for vllm healthchecks. I found this was too short for training cases with long generations.

[gemini-code-assist]

Summary of Changes

Hello @hamishivi, 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 enhances the stability of vLLM engine operations by extending the health check timeout, which is crucial for successful initialization of large language models. Concurrently, it introduces a robust evaluation framework for assessing the performance of tool-use models across multiple benchmarks, significantly improving the project's analytical capabilities for model outputs.

Highlights

• vLLM Health Timeout Increase: The timeout for checking vLLM engines are ready has been increased from 300 seconds (5 minutes) to 600 seconds (10 minutes). This change provides more time for vLLM engine initialization, preventing premature timeouts.
• New Evaluation Script for Tool-Use Models: A new Python script, scripts/eval_results.py, has been added to evaluate inference results from various tool-use models. It supports evaluation for AIME (math verification with majority voting), SimpleQA (LLM-as-judge grading), and GPQA (simple letter comparison).

🧠 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 increases the vLLM health check timeout from 5 to 10 minutes, which is a reasonable change to allow more time for engine initialization. It also introduces a new evaluation script scripts/eval_results.py for AIME, SimpleQA, and GPQA datasets. The script is well-structured and functional. My review includes suggestions to improve code quality by replacing a magic number with a constant in open_instruct/grpo_fast.py and refactoring the new evaluation script to reduce code duplication.