Shawn Im

13 papers A* 2Journal 10Unranked 1
YearRankTypeTitle / Venue / Authors
2026 J jnl
How Do Transformers Learn to Associate Tokens: Gradient Leading Terms Bring Mechanistic Interpretability.
CoRR
Shawn Im, Changdae Oh, Zhen Fang, Sharon Li
2026 J jnl
How Well Can Preference Optimization Generalize Under Noisy Feedback?
Trans. Mach. Learn. Res.
Shawn Im, Yixuan Li
2025 J jnl
A Unified Understanding and Evaluation of Steering Methods.
CoRR
Shawn Im, Yixuan Li
2025 J jnl
How Well Can Preference Optimization Generalize Under Noisy Feedback?
CoRR
Shawn Im, Sharon Li
2025 — conf
Position: Challenges and Future Directions of Data-Centric AI Alignment.
ICML (Position Papers)
Min-Hsuan Yeh, Jeffrey Wang, Xuefeng Du, Seongheon Park, Leitian Tao, Shawn Im, Yixuan Li
2025 J jnl
Towards Interpretability Without Sacrifice: Faithful Dense Layer Decomposition with Mixture of Decoders.
CoRR
James Oldfield, Shawn Im, Yixuan Li, Mihalis A. Nicolaou, Ioannis Patras, Grigorios G. Chrysos
2025 A* conf
Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach.
ICML
Changdae Oh, Zhen Fang, Shawn Im, Xuefeng Du, Yixuan Li
2025 J jnl
Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach.
CoRR
Changdae Oh, Zhen Fang, Shawn Im, Xuefeng Du, Yixuan Li
2025 J jnl
Visual Instruction Bottleneck Tuning.
CoRR
Changdae Oh, Jiatong Li, Shawn Im, Yixuan Li
2024 J jnl
On the Generalization of Preference Learning with DPO.
CoRR
Shawn Im, Yixuan Li
2024 A* conf
Understanding the Learning Dynamics of Alignment with Human Feedback.
ICML
Shawn Im, Yixuan Li
2024 J jnl
Understanding the Learning Dynamics of Alignment with Human Feedback.
CoRR
Shawn Im, Yixuan Li
2023 J jnl
Evaluating the Utility of Model Explanations for Model Development.
CoRR
Shawn Im, Jacob Andreas, Yilun Zhou