Sharon Li

11 papers University of Wisconsin-Madison

• Can DPO Learn Diverse Human Values? A Theoretical Scaling Law
• Clean First, Align Later: Benchmarking Preference Data Cleaning for Reliable LLM Alignment
• Debate or Vote: Which Yields Better Decisions in Multi-Agent Large Language Models?
• GLSim: Detecting Object Hallucinations in LVLMs via Global-Local Similarity
• GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization
• Harnessing Feature Resonance under Arbitrary Target Alignment for Out-of-Distribution Node Detection
• Limited Preference Data? Learning Better Reward Model with Latent Space Synthesis
• MetaMind: Modeling Human Social Thoughts with Metacognitive Multi-Agent Systems
• Towards Interpretability Without Sacrifice: Faithful Dense Layer Decomposition with Mixture of Decoders
• Visual Instruction Bottleneck Tuning
• Your Pre-trained LLM is Secretly an Unsupervised Confidence Calibrator