ReadCtrl: Personalizing text generation with readability-controlled instruction learning
Hieu Tran, Zonghai Yao, Lingxi Li, Hong Yu
- ReadCtrl introduces a dynamic framework that enables LLMs to generate content at various (near continuous) complexity levels, thereby enhancing their versatility across different applications.
- I contributed to the engineering foundation for this research, especially on the human evaluation framework.
Lingxi Li, Zonghai Yao, Sunjae Kwon, Hong Yu
- Efficient, incremental data architecture for taming the full Wikipedia revision history without blowing through storage or query budgets.
- I studied, designed, and implemented the architecture for this entire research, and conducted the verification experiments.
PaniniQA: Enhancing Patient Education Through Interactive Question Answering
Pengshan Cai, Zonghai Yao, Fei Liu, Dakuo Wang, Meghan Reilly, Huixue Zhou, Lingxi Li, Yi Cao, Alok Kapoor, Adarsha Bajracharya, Dan Berlowitz, Hong Yu
- Interactive clinical QA system that helps patients grasp discharge instructions by turning dense documents into conversational answers and reminders.
- I contributed to the engineering foundation for this research, especially the human-in-the-loop evaluation framework.
© 2025 Lingxi Li.
San Francisco