Skip to content

[RadixTree][9/N Refactor]: Support unified init_load_back params#20590

Merged
ispobock merged 4 commits intosgl-project:mainfrom
hzh0425:tree/refactor_init_load_back
Mar 18, 2026
Merged

[RadixTree][9/N Refactor]: Support unified init_load_back params#20590
ispobock merged 4 commits intosgl-project:mainfrom
hzh0425:tree/refactor_init_load_back

Conversation

@hzh0425
Copy link
Copy Markdown
Collaborator

@hzh0425 hzh0425 commented Mar 14, 2026

Motivation

In the future, we may need to add some parameters in init_load_back, such as swa_hit_length、 req (for Mamba).

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

Review Process

  1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
  2. Get approvals from CODEOWNERS and other reviewers.
  3. Trigger CI tests with comments or contact authorized users to do so.
    • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
  4. After green CI and required approvals, ask Merge Oncalls to merge.

@gemini-code-assist
Copy link
Copy Markdown
Contributor

Summary of Changes

Hello, 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 refactors the init_load_back function by unifying its parameters into a single InitLoadBackParams dataclass. This change enhances the extensibility and maintainability of the cache loading mechanism, allowing for easier addition of new parameters in the future without modifying multiple function signatures. It streamlines the interface for preparing KV cache loading from host to device across different cache types.

Highlights

  • Unified Parameters for init_load_back: A new InitLoadBackParams dataclass was introduced to centralize and standardize the parameters passed to the init_load_back function across different cache implementations.
  • Refactored Method Signatures: The init_load_back method signatures in BasePrefixCache, HiMambaRadixCache, HiRadixCache, and SessionAwareCache were updated to accept a single InitLoadBackParams object instead of multiple individual arguments.
  • Updated Call Site: The call to self.tree_cache.init_load_back in schedule_policy.py was modified to construct and pass the new InitLoadBackParams object, ensuring consistency with the refactored method signatures.

🧠 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.

Changelog
  • python/sglang/srt/managers/schedule_policy.py
    • Imported InitLoadBackParams from base_prefix_cache.
    • Modified the call to self.tree_cache.init_load_back to pass an instance of InitLoadBackParams instead of separate arguments.
  • python/sglang/srt/mem_cache/base_prefix_cache.py
    • Defined the InitLoadBackParams dataclass, including last_host_node, host_hit_length, and optional mem_quota and req fields.
    • Updated the init_load_back method signature to accept a single params: InitLoadBackParams argument.
  • python/sglang/srt/mem_cache/hi_mamba_radix_cache.py
    • Imported InitLoadBackParams.
    • Adjusted the init_load_back method signature to accept params: InitLoadBackParams and extracted last_node and mem_quota from it.
  • python/sglang/srt/mem_cache/hiradix_cache.py
    • Imported InitLoadBackParams.
    • Adjusted the init_load_back method signature to accept params: InitLoadBackParams and extracted last_node and mem_quota from it, removing the unused host_hit_length parameter.
  • python/sglang/srt/mem_cache/session_aware_cache.py
    • Imported InitLoadBackParams.
    • Updated the init_load_back method signature to accept params: InitLoadBackParams and passed this object directly to the inner cache's init_load_back method.
Activity
  • The pull request was opened by hzh0425.
  • No human activity (comments, reviews) has been recorded yet.
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.

Copy link
Copy Markdown
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 refactors the init_load_back method across several cache implementations by introducing a new InitLoadBackParams dataclass. This dataclass unifies the parameters (last_host_node, host_hit_length, mem_quota, req) for the init_load_back method, simplifying its signature in BasePrefixCache, HiMambaRadixCache, HiRadixCache, and SessionAwareCache. The schedule_policy.py file is updated to call init_load_back using an instance of this new parameter object.

hzh0425 added 2 commits March 16, 2026 13:54
…ad_back2

# Conflicts:
#	python/sglang/srt/mem_cache/base_prefix_cache.py
#	python/sglang/srt/mem_cache/hi_mamba_radix_cache.py
#	python/sglang/srt/mem_cache/hiradix_cache.py
#	python/sglang/srt/mem_cache/session_aware_cache.py
@ispobock ispobock merged commit c43d495 into sgl-project:main Mar 18, 2026
201 of 214 checks passed
JustinTong0323 pushed a commit to JustinTong0323/sglang that referenced this pull request Apr 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants