[Fix] Prevent client from importing ray via twinkle.server.common.serialize

[xichengpro]

When I run python cookbook/client/twinkle/self_host/grpo.py, I get an error.

Traceback (most recent call last):
  File "/Users/dubingnan/workCode/twinkle/cookbook/client/twinkle/self_host/grpo.py", line 295, in <module>
    train()
  File "/Users/dubingnan/workCode/twinkle/cookbook/client/twinkle/self_host/grpo.py", line 120, in train
    dataset = create_gsm8k_dataset()
              ^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/dubingnan/workCode/twinkle/cookbook/client/twinkle/self_host/grpo.py", line 94, in create_gsm8k_dataset
    dataset = Dataset(DatasetMeta('/data/dubingnan/dbn-ceph/datasets/data/gsm8k/train.jsonl', subset_name='default', split='train'))
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/dubingnan/workCode/twinkle/src/twinkle_client/dataset/base.py", line 27, in __init__
    response = http_post(
               ^^^^^^^^^^
  File "/Users/dubingnan/workCode/twinkle/src/twinkle_client/http/http_utils.py", line 147, in http_post
    serialized_json = _serialize_params(json_data)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/dubingnan/workCode/twinkle/src/twinkle_client/http/http_utils.py", line 48, in _serialize_params
    from twinkle.server.common.serialize import serialize_object
  File "/Users/dubingnan/workCode/twinkle/src/twinkle/server/common/__init__.py", line 4, in <module>
    from .router import StickyLoraRequestRouter
  File "/Users/dubingnan/workCode/twinkle/src/twinkle/server/common/router.py", line 3, in <module>
    from ray.serve.request_router import (FIFOMixin, MultiplexMixin, PendingRequest, ReplicaID, ReplicaResult,
ModuleNotFoundError: No module named 'ray'

When client code (e.g., grpo.py ) calls Dataset() to create a dataset, it eventually invokes _serialize_params() in http_utils.py to serialize parameters. When this function encounters an object with dict (such as DatasetMeta ), it executes:
from twinkle.server.common.serialize import serialize_object
However, Python executes twinkle.server.common.init.py first when importing twinkle.server.common.serialize . This init.py imports StickyLoraRequestRouter (from router.py ), which further imports ray.serve . As a result, even though the client code doesn't need ray at all, the import chain causes ModuleNotFoundError: No module named 'ray' .

PR type

• Bug Fix
• New Feature
• Document Updates
• More Models or Datasets Support

PR information

Create a standalone serialize.py module under twinkle_client/common/ so that the client uses its own serialization logic without depending on the server package structure.

Experiment results

Paste your experiment result here(if needed).

[gemini-code-assist]

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 resolves a critical ModuleNotFoundError that occurred when client applications, such as grpo.py, attempted to serialize objects. The root cause was an indirect dependency on the ray library, which was pulled in via a server-side serialization utility. The solution involves refactoring the serialization logic into a dedicated client-side module, thereby isolating client code from server-specific dependencies and ensuring that client applications can operate without requiring server-only packages.

Highlights

• Resolved ModuleNotFoundError for ray: Addressed an issue where client-side code (e.g., grpo.py) indirectly attempted to import the ray module through twinkle.server.common.serialize, leading to a ModuleNotFoundError because ray is not a client dependency.
• Introduced dedicated client serialization module: Created a new module, src/twinkle_client/common/serialize.py, to house serialization logic specifically for client-side operations, preventing unintended dependencies on server-side packages.
• Refactored serialization import path: Updated src/twinkle_client/http/http_utils.py to import the serialize_object function from the newly created client-specific serialization module, ensuring proper dependency isolation.

🧠 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.

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.

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 effectively resolves the ModuleNotFoundError by decoupling the client-side serialization logic from the server-side implementation. Creating a dedicated serialize.py for the client is a good approach to prevent pulling in server-side dependencies like ray. I've found a couple of potential issues in the newly added serialization file that could lead to runtime errors or confusion, and I've left specific comments with suggestions for improvement.

[Yunnglin]

LGTM! Thanks for the contribution. Merging now.