Shawn Im
Hello! I am a PhD student at UW-Madison working with Prof. Sharon Li. I am thankful to be supported by the NSF GRFP. Currently, I am interning at Apple MLR. Previously, I was an undergraduate at MIT in mathematics and computer science. My interest is in building methods for reliable understanding and control of the behavior of machine learning models, especially in the context of safety. I see model reliability not only as a way of ensuring models are beneficial but also as a key part of working towards more complex systems. Currently, I am working on
- AI Safety
- Theory of ML
- Interpretability
Research
Towards Interpretability Without Sacrifice: Faithful Dense Layer Decomposition with Mixture of Decoders
James Oldfield, Shawn Im, Yixuan Li, Mihalis A. Nicolaou, Ioannis Patras, Grigorios G Chrysos
Preprint, 2025
[Paper]
Position: Challenges and Future Directions of Data-Centric AI Alignment
Min-Hsuan Yeh, Jeffrey Wang, Xuefeng Du, Seongheon Park, Leitian Tao, Shawn Im, Yixuan Li
In Proceedings of International Conference on Machine Learning (ICML), 2025
[Paper]
Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach
Changdae Oh, Zhen Fang, Shawn Im, Xuefeng Du, Yixuan Li
In Proceedings of International Conference on Machine Learning (ICML), 2025
[Paper]
A Unified Understanding and Evaluation of Steering Methods
Shawn Im, Yixuan Li
Preprint, 2025
[Paper]
On the Generalization of Preference Learning with DPO
Shawn Im, Yixuan Li
Preprint, 2024
[Paper]
Evaluating the Utility of Model Explanations for Model Development
Shawn Im, Jacob Andreas, Yilun Zhou
NeurIPS Workshop on Attributing Model Behavior at Scale (ATTRIB), 2023
[Paper]
Other