Add learn_to_ask example

[chenyushuo]

Description

As the title says.

Checklist

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

• Code has passed all tests
• Docstrings have been added/updated in Google Style
• Documentation has been updated
• Code is ready for review

[gemini-code-assist]

Summary of Changes

Hello @chenyushuo, 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 adds a new learn2ask example, establishing a complete pipeline for developing and evaluating a medical assistant large language model. The example covers data preparation, including session segmentation and information extraction, model training using reinforcement learning with a custom reward function, and a robust evaluation process to assess the model's inquiry capabilities.

Highlights

• New Example: learn2ask: Introduces a comprehensive learn2ask example focused on training a medical assistant LLM to conduct effective medical inquiries.
• Data Processing Pipeline: Includes scripts for preparing medical dialogue data, segmenting sessions into context-future pairs, and extracting ground-truth labels for reward calculation.
• LLM-powered Information Extraction: Features a module for extracting clinical attributes from patient-doctor dialogues using LLMs, supporting both online API and local vLLM inference.
• Reinforcement Learning Workflow: Defines a specialized workflow (Learn2AskWorkflow) that integrates different system prompts and a sophisticated reward function for training the medical assistant model.
• Evaluation Framework: Provides scripts for generating rollout samples using vLLM and evaluating model outputs with a separate grader LLM (e.g., qwen2.5-32b-instruct).

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 learn2ask example, including data preparation scripts, workflow logic, prompts, and configuration. The implementation is comprehensive, but there are several areas for improvement. I've identified a critical bug in the evaluation script that will cause a KeyError, and some high-severity issues related to data processing robustness, scalability, and use of global variables in the workflow. I've also included several medium-severity suggestions to improve code quality, maintainability, and correctness across the new files. My feedback includes code suggestions to address these points.

[gemini-code-assist]

Warning

Gemini encountered an error creating the review. You can try again by commenting /gemini review.

[yxdyc]

The directory name should be consistent with arXiv version: examples/learn_to_ask.

[chenyushuo]

/gemini review

[gemini-code-assist]

Code Review

This pull request adds the learn2ask example, including data preparation scripts, workflow definitions, and configuration files. While the example is comprehensive, there are several critical issues that will prevent it from running correctly. These include bugs that will cause crashes due to incorrect data handling and inconsistent data formats between scripts, as well as a major performance bottleneck from repeatedly initializing a large language model inside a loop. Additionally, there are several high-severity issues such as broken links and incorrect paths in the documentation, and invalid model names in the configuration. I've also pointed out some medium-severity issues related to code quality, like the use of global state, code duplication, and redundant method calls. Addressing these points will be crucial for making the example functional, performant, and maintainable.