Skip to content

test: support return logprobs in bench_offline_throughput test #12462

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 2 commits into from
Nov 4, 2025
Merged

test: support return logprobs in bench_offline_throughput test #12462

merged 2 commits into from
Nov 4, 2025

Conversation

@aftersnow
Copy link
Contributor

@aftersnow aftersnow commented Oct 31, 2025

Motivation

This PR adds support for returning log probabilities in the bench_offline_throughput test script. Currently, the benchmark script does not provide an option to retrieve logprobs during throughput testing, which limits its ability to test and benchmark the logprob functionality. This enhancement enables users to test and validate logprob generation performance in offline throughput scenarios.

Modifications

  1. Added new command-line arguments to BenchArgs:

    • --return-logprob: A boolean flag to enable logprob return
    • --logprob-start-len: An integer parameter to control the starting position for logprob calculation
      • Default value is -1 (only return logprobs for output tokens)
      • Set to 0 to return logprobs for all tokens including input

  2. Updated BenchArgs dataclass:

    • Added return_logprob field (default: False)
    • Added logprob_start_len field (default: -1)

  3. Modified throughput_test_once function:

    • Added return_logprob and logprob_start_len parameters
    • Updated backend.generate() calls to pass these parameters

  4. Updated throughput_test function:

    • Passed the new logprob parameters to both warmup and actual benchmark runs

Accuracy Tests

Benchmarking and Profiling

Checklist

@gemini-code-assist
Copy link
Contributor

gemini-code-assist bot commented Oct 31, 2025

Summary of Changes

Hello @aftersnow, 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 bench_offline_throughput test script by integrating support for log probability generation. It introduces new command-line arguments and updates the core benchmarking functions to allow users to test and validate the performance of logprob functionality in offline throughput scenarios, which was previously not possible. This change enables a more comprehensive evaluation of model capabilities under various conditions.

Highlights

  • New CLI Arguments: Introduced --return-logprob (boolean flag) and --logprob-start-len (integer, default -1) to the bench_offline_throughput script for controlling log probability generation. The logprob_start_len parameter allows specifying whether to return logprobs for only output tokens (-1) or all tokens including input (0).
  • BenchArgs Update: The BenchArgs dataclass now includes return_logprob (default False) and logprob_start_len (default -1) to store the new command-line arguments, ensuring they are properly managed within the benchmark configuration.
  • Logprob Parameter Propagation: The throughput_test_once function was updated to accept return_logprob and logprob_start_len as parameters and pass them directly to the backend.generate() calls, enabling the backend to handle log probability requests.
  • Benchmark Integration: The throughput_test function now passes the newly introduced logprob parameters to both the warmup and actual benchmark runs of throughput_test_once, ensuring that logprob functionality can be thoroughly tested and benchmarked.
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[bot]
gemini-code-assist bot reviewed Oct 31, 2025
View reviewed changes
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request adds support for returning log probabilities in the bench_offline_throughput.py script. The changes are well-structured, adding new arguments to BenchArgs and propagating them correctly to the backend. My feedback includes a minor suggestion to improve the clarity of a command-line argument's help message.

@aftersnow @gemini-code-assist
Update python/sglang/bench_offline_throughput.py
786c586 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
@Kangyan-Zhou Kangyan-Zhou merged commit 0b24af4 into sgl-project:main Nov 4, 2025
41 of 44 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@zhaochenyang20 zhaochenyang20 zhaochenyang20 approved these changes

@merrymercy merrymercy Awaiting requested review from merrymercy merrymercy is a code owner

@Ying1123 Ying1123 Awaiting requested review from Ying1123 Ying1123 is a code owner

@zhyncs zhyncs Awaiting requested review from zhyncs

@hnyls2002 hnyls2002 Awaiting requested review from hnyls2002 hnyls2002 is a code owner

+1 more reviewer

@gemini-code-assist gemini-code-assist[bot] gemini-code-assist[bot] left review comments

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@aftersnow @zhaochenyang20 @Kangyan-Zhou