Andrew (Ching-Yuan) Bai

Hello! Welcome to my personal website.

I am a 5th year PhD student in Computer Science at UCLA, where I work with Cho-Jui Hsieh. My research interest involves studying the memorization and forgetting mechanisms when training machine learning models.

Since the LLM boom, I focus on applications in LLM post-training. My latest project investigates why fine-tuning LLM with RLHF leads to less forgetting compared to supervised fine-tuning. Previously I worked on selecting the best instruction-tuning checkpoint for downstream preference alignment (RLHF) performance in LLMs. I also investigated the forgetting of context-awareness when pretrained LLMs are supervised fine-tuned on instruction data.

My favorite part of research is collaborating and engaging in insightful discussions with fellow reseachers. I have a diverse exposure to various topics through collaborations. My collaboration projects include reward modeling, long video generation, LLM agents, instruction selection, diffusion model data memorization, and prompt optimization.

Before the LLM boom I worked on data selection. I proposed a simple and easy-to-adopt sample selection schemes for prioritizing data samples during training. I also worked on developing practical interpretation methods for black-box models, allowing humans to better understand and trust machine learning models in real life (concept-based interpretability).

I was an undergraduate student in Computer Science at National Taiwan University. I worked with Hsuan-Tien Lin on generative modeling and time series forecasting. We held the first-ever generative modeling competition in collaboration with Kaggle. I also worked with Chung-Wei Lin on system verification and falsification.

Email: andrewbai [AT] cs.ucla.edu

Links: [Resume] [CV] [Github] [Google Scholar] [Linkedin]

Publications

2025

2024

2023

2022

2021

2020