Add README (Chinese) for tuner

[hiyuchang]

📝 PR Type

• Add new sample
• Update existing sample
• Add new test cases
• Fix test failures
• Documentation/Configuration update

---

📚 Description

[Please briefly describe the background, changes, and purpose of this PR. For example:

• Added game_werewolves to demonstrate XYZ functionality in agentscope.
• Fixed test failures in game_test.py caused by agentscope interface changes.
• Updated dependency installation instructions in README.md of agentscope-samples.]

---

🧪 Testing Validation

[Please explain how to validate the changes:

1. How to run the added/modified test cases?
2. Is integration testing with agentscope required?
3. Has code been formatted (e.g., pre-commit)?]

---

✅ Checklist

Please complete the following checks before submitting the PR:

• All sample code has been formatted with pre-commit run --all-files
• All new/modified test cases have passed (run pytest tests/)
• Test coverage has not decreased (if applicable)
• Sample code follows agentscope best practices (e.g., config management, logging)
• Related documentation in agentscope-samples has been updated (e.g., README.md)

[gemini-code-assist]

Summary of Changes

Hello @hiyuchang, 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 significantly enhances the tuner examples by introducing a new data-centric training example for math agents, complete with a difficulty_based task selector. It also greatly improves the documentation by adding Chinese translations for several key examples and standardizing image rendering across existing README files. Furthermore, it refines the learn_to_ask example's reward logic and provides a more in-depth qualitative analysis for the werewolves game example, making the examples more robust, accessible, and informative.

Highlights

• New Data Augmentation Example: A new example (tuner/data_augment) has been introduced, demonstrating how to train math agents using data-centric strategies, specifically a difficulty_based task selector for improved training efficiency.
• Multilingual Documentation Expansion: Comprehensive Chinese README_zh.md files have been added for the new data_augment example and existing learn_to_ask and werewolves examples, significantly enhancing accessibility for Chinese-speaking users.
• Documentation Improvements: Image embedding in several README.md files (email_search, frozen_lake, learn_to_ask, math_agent, werewolves) has been standardized using div align="center" for better rendering and presentation.
• Code Refinement in Learn2Ask: The reward calculation logic within the learn_to_ask judge function has been simplified, making it more robust and readable by removing an unnecessary assertion and clarifying the reward combination.
• Enhanced Werewolves Example Details: The werewolves example's README.md has been expanded with more precise descriptions of game mechanics, algorithm parameters, and a detailed qualitative analysis of agent behavior before and after training, including concrete examples of strategic play.
• Terminology Update: Project terminology across several README.md files has been standardized, replacing "AS-tune" with "AgentScope-Tuner" for consistency.

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 a new data_augment example for the tuner, including English and Chinese READMEs. It also adds Chinese READMEs for the learn_to_ask and werewolves examples, and makes some formatting improvements to existing documentation.

The new data_augment example is well-structured. However, I've found a few issues:

• The configuration is defined in both the Python script and the YAML files, which is redundant and confusing. It would be better to rely on the YAML files as the single source of truth for configuration.
• The READMEs for data_augment and learn_to_ask contain placeholder links to arXiv papers that should be corrected.
• There's a discrepancy in the learn_to_ask READMEs where the code snippet for learn2ask_judge is a refactored, cleaner version that isn't reflected in the actual main.py file. It would be great to update the code to match the improved version in the documentation.
• Some new YAML files are missing a final newline character.

Overall, this is a great contribution that improves the documentation and examples for the tuner. Addressing the points above will enhance the clarity and maintainability of the new example.