Lee Lab @ UW Madison

Research focus:
(1) understanding and improving foundation models
(2) efficient adaptation of foundation models

Research Highlights

• Arxiv’25 / In-Context Learning with Hypothesis-Class Guidance
• Arxiv’25 / LLM-Lasso: A Robust Framework for Domain-Informed Feature Selection and Regularization
• Arxiv’25 / VersaPRM: Multi-Domain Process Reward Model via Synthetic Reasoning Data
• Arxiv’25 / Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challenges
• ICLR’25 / Looped Transformers for Length Generalization / Summary / Github
• ICLR’25 / From Artificial Needles to Real Haystacks: Improving Retrieval Capabilities in LLMs by Finetuning on Synthetic Data / Summary / Github
• ICLR’25 / Rare-to-Frequent: Unlocking Compositional Generation Power of Diffusion Models on Rare Concepts with LLM Guidance / Summary / Github
• Arxiv’24 / Parameter-Efficient Fine-Tuning of State Space Models / Summary / Github
• Arxiv’24 / Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition / Summary
• Arxiv’24 / ENTP: Encoder-only Next Token Prediction / Summary / Github
• ICML’24 / Dual operating modes of ICL / Summary / Github
• ICML’24 / Mamba + Transformer / Github
• ICLR’24 / Expressive power of LoRA / Summary / Github
• ICLR’24 / Teaching arithmetic to small TFs / Summary / Github
• ICLR’24 / Looepd Transformer for meta ICL / Summary / Github
• NeurIPS’23 / RL for (text-to-image) diffusion model finetuning / Summary / Github
• ICML’23 / RL for diffusion model finetuning / Summary / Github
• NeurIPS’22 / Finetuning LLMs on non-textual data / Summary / Github

We sincerely appreciate the support provided by our sponsors: NSF, Amazon, and FuriosaAI.

Openings

• [postdocs] We are looking for a postdoc interested in the theoretical and algorithmic aspects of foundation models, particularly LLMs. If you want to work with us, please email me your CV and a research statement. I strongly recommend reading our lab’s recent papers.
• [PhD students] We are looking for PhD students (ECE or CS) interested in the theoretical and algorithmic aspects of foundation models, particularly LLMs. If you want to work with us, please email me your CV and a research statement. I strongly recommend reading our lab’s recent papers.
• [MS students] I am not currently looking for MS students.
• [undergraduate students] You might want to consider taking my class first. I teach various machine learning courses (ECE 532, 539, 561, 570, 761, …).

News

• (Jan. 2025) [ICLR’25] Looped Transformers for Length Generalization / Summary
• (Jan. 2025) [ICLR’25] From Artificial Needles to Real Haystacks: Improving Retrieval Capabilities in LLMs by Finetuning on Synthetic Data / Summary
• (Jan. 2025) [ICLR’25] Rare-to-Frequent: Unlocking Compositional Generation Power of Diffusion Models on Rare Concepts with LLM Guidance / Summary
• (Jan. 2025) [TMLR’25] Improving CLIP Counting Accuracy via Parameter-Efficient Fine-Tuning
• (Jan. 2025) [TMLR’25] Buffer-based Gradient Projection for Continual Federated Learning
• (Nov. 2024) Dr. Jungtaek Kim joined our lab as a postdoc! Welcome!
• (July 2024) [COLM’24] Can MLLMs Perform Text-to-Image In-Context Learning?
• (June 2024) [TMLR’24] Mini-Batch Optimization of Contrastive Loss
• (April 2024) [UAI’24] Memorization capacity of additive fine-Tuning
• (Apr. 2024) Amazon Research Awards
  Our group will develop principled approaches to prompt engineering through the information/coding-theoretic lens. Huge thanks to Amazon and my amazing collaborators and students!
• (Mar. 2024) NSF CAREER Award
  Our group will develop a unified theory and new algorithms with provable guarantees for learning with frozen pretrained models, also known as foundation models. Huge thanks to NSF and my amazing collaborators and students!

Selected Talks on Foundation Models

• (Nov. 2024) Seminars on AI Core and Applications, Seoul National University
• (Oct. 2024) Mathematical Principles in Foundation Models, 2024 SIAM Conference on Mathematics of Data Science
  Title: Dual Operating Modes of ICL
• (Apr. 2024) The Johns Hopkins University CIS/MINDS seminar
  Title: Theoretical Exploration of Foundation Model Adaptation Methods
• (Mar. 2024) The 58th Annual Conference on Information Sciences and Systems @ Princeton University
  Title: A Probabilistic Framework for Understanding In-Context Task Learning and Retrieval
• (Feb. 2024) 2024 Information Theory and Applications Workshop
  Title: The Expressive Power of Low-Rank Adaptation (LoRA)
• (Feb. 2024) Foundations of Data Science - Virtual Talk Series @ UCSD/NSF TRIPODS Institute on Emerging CORE Methods in Data Science (EnCORE)
  Title: Theoretical Exploration of Foundation Model Adaptation Methods (video)
• (Dec. 2023) CSP Seminar @ University of Michigan
  Title: Towards a Theoretical Understanding of Parameter-Efficient Fine-Tuning (and Beyond) (video)
• (Nov. 2023) Efficient ML workshop @ Google Research New York
  Title: The Expressive Power of Low-Rank Adaptation (LoRA)