feat: preview filename from tuning_fused_moe_triton.py

[lianakoleva]

Motivation

Since the tuning_fused_moe_triton.py script may be used when working backwards from the following warning, and the tuning process can take some time, it would be helpful to preview the filename of the generated config to ensure that the desired one is being created:

[2025-10-28 09:42:18 TP0] Using default MoE kernel config. Performance might be sub-optimal! Config file not found at /sgl-workspace/sglang/python/sglang/srt/layers/moe/fused_moe_triton/configs/triton_3_4_0/E=60,N=704,device_name=NVIDIA_B200,dtype=fp8_w8a8.json, you can create them with https://github.com/sgl-project/sglang/tree/main/benchmark/kernels/fused_moe_triton

Modifications

The filename now displays before the tuning begins:

~/sglang/benchmark/kernels/fused_moe_triton python tuning_fused_moe_triton.py --model Qwen/Qwen1.5-MoE-A2.7B --dtype fp8_w8a8 --tune 
Namespace(model='Qwen/Qwen1.5-MoE-A2.7B', tp_size=2, dtype='fp8_w8a8', per_channel_quant=False, seed=0, batch_size=None, tune=True, disable_shared_experts_fusion=False)
- Start tuning over 1920 configurations...
+ Start tuning over 1920 configurations to create E=60,N=704,device_name=NVIDIA_B200,dtype=fp8_w8a8.json...

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 @lianakoleva, 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 user experience of the tuning_fused_moe_triton.py script by providing an early preview of the output configuration filename. This allows users to confirm the target file before initiating a potentially time-consuming tuning process, addressing a common pain point where the tuning process might complete only to reveal an incorrectly named output file.

Highlights

• User Experience Improvement: The tuning_fused_moe_triton.py script now previews the generated configuration filename before the tuning process begins, allowing users to confirm the output file early.
• Code Refactoring: The logic for generating the configuration filename has been extracted into a new get_filename function, separating it from the save_configs function, which now accepts the pre-generated filename.

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 refactors the tuning_fused_moe_triton.py script to preview the output configuration filename before starting the potentially long-running tuning process. The change is a good improvement for user experience. The refactoring splits the filename generation logic from the file saving logic into a new get_filename function. My review includes suggestions to improve the signature and type hinting of the new function for better code clarity and correctness.

[gemini-code-assist]

To align with the suggested changes to the get_filename function signature, the unused arguments hidden_size and topk should be removed from this call.

        filename = get_filename(
            E,
            shard_intermediate_size,
            dtype,
            use_fp8_w8a8,
            use_int8_w8a8,
            use_int8_w8a16,
            per_channel_quant,
            block_shape,
        )

[lianakoleva]

Happy to make this change if desired, I just kept it close to the original

[BBuf]

LGTM.