Add examples for frozenlake and emailsearch

[hiyuchang]

📝 PR Type

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

---

📚 Description

Add examples for frozenlake and emailsearch to show the usage of Agentscope-Tuner.

---

🧪 Testing Validation

Follow the README to run the examples.

---

✅ 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)

[cla-assistant]

All committers have signed the CLA.

[Copilot]

Pull request overview

This PR adds two comprehensive examples demonstrating AgentScope-Tuner usage for reinforcement learning fine-tuning: FrozenLake (a grid navigation task) and Email Search (an information retrieval task). Both examples showcase multi-step agent-environment interactions with proper documentation and configuration.

Changes:

• Added FrozenLake example with agent, environment, utilities, data generation, configuration, and bilingual documentation
• Added Email Search example with agent, tools, database setup, judge function, configuration, and bilingual documentation
• Included training result visualizations (PNG files) for both examples

Reviewed changes

Copilot reviewed 15 out of 18 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tuner/frozen_lake/main.py	Workflow function implementing multi-step agent-environment interaction for FrozenLake navigation
tuner/frozen_lake/get_frozenlake_data.py	Data generation script creating training/test datasets with random map configurations
tuner/frozen_lake/_frozenlake_env.py	Environment wrapper extending Gymnasium's FrozenLake with custom rendering and action handling
tuner/frozen_lake/_frozenlake_agent.py	ReActAgent implementation for FrozenLake with action parsing and state management
tuner/frozen_lake/_utils.py	Utility functions including map generation, validation, and prompt templates
tuner/frozen_lake/config.yaml	Training configuration with GRPO algorithm settings and model parameters
tuner/frozen_lake/README.md	English documentation explaining task, setup, and execution
tuner/frozen_lake/README_zh.md	Chinese documentation explaining task, setup, and execution
tuner/frozen_lake/critic_rewards_mean.png	Training reward visualization
tuner/email_search/main.py	Workflow and judge functions with LLM-as-a-Judge for email search tasks
tuner/email_search/prepare_data.py	Database setup script downloading and processing Enron email dataset
tuner/email_search/_email_search_agent.py	ReActAgent with email search and read tools
tuner/email_search/_utils.py	Tool implementations for searching/reading emails and data models
tuner/email_search/config.yaml	Training configuration with auxiliary judge model settings
tuner/email_search/README.md	English documentation with task description and example outputs
tuner/email_search/README_zh.md	Chinese documentation with task description and example outputs

Comments suppressed due to low confidence (1)

tuner/email_search/prepare_data.py:1

• The path construction uses multiple '..' which makes it fragile and unclear. Consider using a more robust approach with os.path.normpath() or defining a clear base directory constant to improve code clarity and maintainability.

# -*- coding: utf-8 -*-

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.