documented sampling.cuh

[yonikremer]

📌 Description

Hey, I had a hard time figuring out what is going on in sampling.cuh, after I figured it out I documented it.
Yes an AI helped my and yes I read the documents and they are good in my opinion.

🔍 Related Issues

🚀 Pull Request Checklist

Thank you for contributing to FlashInfer! Before we review your pull request, please make sure the following items are complete.

✅ Pre-commit Checks

• [ V ] I have installed pre-commit by running pip install pre-commit (or used your preferred method).
• [ V ] I have installed the hooks with pre-commit install.
• [ V ] I have run the hooks manually with pre-commit run --all-files and fixed any reported issues.

If you are unsure about how to set up pre-commit, see the pre-commit documentation.

🧪 Tests

• [ X ] Tests have been added or updated as needed. - Only document chnages, no need to change the tests
• [ V ] All tests are passing (unittest, etc.).

Reviewer Notes

[gemini-code-assist]

Summary of Changes

Hello @yonikremer, 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 focuses on significantly improving the documentation within the include/flashinfer/sampling.cuh file. It introduces a comprehensive file-level overview and detailed Doxygen comments for nearly all CUDA kernels and host utilities related to softmax and token sampling. The goal is to enhance the clarity and understanding of the complex algorithms implemented, making the codebase more accessible and maintainable for current and future developers.

Highlights

• Comprehensive File Documentation: A detailed Doxygen comment has been added to the sampling.cuh header, providing a high-level overview of its purpose, design principles, threading model, error handling, and usage for softmax and token sampling.
• Function-Level Documentation: Extensive Doxygen comments have been introduced for numerous CUDA kernels and host utilities within sampling.cuh, including GetMinMaxValue, OnlineSoftmaxFusedKernel, SamplingFromLogitsKernel, TopKSamplingFromProbKernel, and OnlineSoftmax.
• Algorithmic Clarity: Inline comments have been added to clarify complex algorithmic details, such as the online merge rule for softmax, Gumbel-max sampling logic, CDF search mechanisms, and iterative pivot refinement in Top-K and Top-P sampling.
• Improved Readability: The added documentation significantly enhances the readability and maintainability of the sampling.cuh file, making it easier for developers to understand the intricate logic behind the high-performance CUDA primitives.

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 comprehensive Doxygen-style documentation to sampling.cuh, significantly improving the readability and maintainability of the CUDA kernels and host utilities for token sampling. The added comments are clear, detailed, and accurately describe the functionality, design, and algorithms used. I have one minor suggestion to improve the clarity of the file-level documentation.

[gemini-code-assist]

The term "in-place" can be ambiguous here. While it's possible for the output buffer to be the same as the logits buffer, the function signature OnlineSoftmax(DType* logits, DType* output, ...) suggests they can be separate. Using "in-place" might imply that the logits buffer is always modified directly, which isn't necessarily true. To improve clarity, I suggest removing "in-place".

 * - Call OnlineSoftmax to convert logits to probabilities into an output buffer.