feat: update ChatRLDataset class and judge model grpo training document

[jc200808]

OpenJudge Version

[The version of OpenJudge you are working on, e.g. import openjudge; print(openjudge.__version__)]

Description

[Please describe the background, purpose, changes made, and how to test this PR]

Checklist

Please check the following items before code is ready to be reviewed.

• Code has been formatted with pre-commit run --all-files command
• All tests are passing
• Docstrings are in Google style
• Related documentation has been updated (e.g. links, examples, etc.)
• Code is ready for review

[gemini-code-assist]

Summary of Changes

Hello @jc200808, 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 clarity and robustness of the GRPO training process. It refines the documentation for Group Relative Policy Optimization, making it easier to understand the training objectives and modes. Concurrently, it improves the ChatRLDataset class by introducing safer data handling through type checks, streamlining the serialization process, and providing more informative error messages during ground truth extraction. These changes collectively contribute to a more stable, maintainable, and user-friendly experience for training judge models.

Highlights

• Documentation Clarity: The GRPO training documentation (README.md) has been significantly improved with clearer formatting, bold text for key terms, expanded tips, and better-structured explanations for the training objective and modes.
• Reward Calculation Description Refinement: The README.md now consolidates the reward conditions for 'Wrong prediction' and 'Failed to extract preference' into a single, clearer statement.
• Robustness in ChatRLDataset: Added a type check (isinstance(msg, dict)) in the _extract_prompt method of chat_rl_dataset.py to ensure safe access to message attributes, preventing potential errors with non-dictionary message types.
• Simplified Serialization Logic: The __getstate__ method in ChatRLDataset has been refactored to use state.pop('dataframe', None), providing a cleaner and more robust way to handle dataframe exclusion during serialization, regardless of a serialize_dataset flag.
• Corrected Enumeration Start Index: The _format_template method in ChatRLDataset now correctly starts the enumeration of principles from 1 using enumerate(principles, start=1).
• Enhanced Ground Truth Extraction Error Handling: The _extract_ground_truth method in ChatRLDataset now initializes helpfulness to 0 and includes more specific error handling by catching Exception as e and printing the error, making debugging easier.

🧠 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.

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 introduces several improvements to the GRPO training components. The documentation in grpo/README.md has been reformatted for better readability. In chat_rl_dataset.py, the ChatRLDataset classes have been updated for robustness and clarity. Key changes include adding type checks when parsing message inputs, simplifying the __getstate__ method to improve checkpointing behavior by excluding the large dataframe, and refactoring the _extract_ground_truth method in PointwiseChatRLDataset for cleaner logic and better error handling. The code changes are well-structured and enhance the maintainability of the dataset handling. I have one suggestion regarding the use of print for logging errors.