Support LoRA in bench_serving oai interface

[lifuhuang]

Motivation

Enable bench_serving to benchmark LoRA inference in SGLang and vLLM.

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.

[gemini-code-assist]

Summary of Changes

Hello @lifuhuang, 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_serving tool by integrating support for benchmarking LoRA inference. The primary goal is to enable performance evaluation of LoRA models across both SGLang and vLLM environments. This is achieved by standardizing how LoRA model identifiers are included in API requests, ensuring seamless compatibility and accurate benchmarking results regardless of the serving backend used.

Highlights

• LoRA Benchmarking Support: Introduced the capability to benchmark LoRA (Low-Rank Adaptation) inference within the bench_serving utility.
• Cross-Backend LoRA Compatibility: Implemented a compatibility layer to handle differing LoRA naming conventions between SGLang and vLLM serving backends.
• API Request Modification: Modified async_request_openai_completions and async_request_openai_chat_completions functions to correctly pass LoRA model names in the request payload.

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 adds support for benchmarking LoRA inference in bench_serving.py for both SGLang and vLLM. The changes introduce logic to handle the different LoRA conventions between the two frameworks by setting both model and lora_path in the request payload when a LoRA name is provided.

My main feedback is to address the code duplication for this new LoRA handling logic, which appears in both async_request_openai_completions and async_request_openai_chat_completions. Extracting this logic into a shared helper function would improve maintainability.

[gemini-code-assist]

This logic for handling LoRA parameters is duplicated in async_request_openai_chat_completions at lines 335-338. To improve maintainability and avoid potential inconsistencies, consider extracting this into a helper function.

For example:

def _add_lora_to_payload(payload: Dict[str, Any], lora_name: Optional[str]):
    """Adds LoRA parameters to the payload for SGLang and vLLM compatibility."""
    if lora_name:
        # Accommodate different LoRA conventions between SGLang and vLLM.
        payload["model"] = lora_name
        payload["lora_path"] = lora_name

You could then replace this block with _add_lora_to_payload(payload, request_func_input.lora_name) in both places.

[Fridge003]

LGTM