Licong Lin

Improved Scaling Laws in Linear Regression via Data Reuse

Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning

A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics

Improving LLM Safety Alignment with Dual-Objective Optimization

Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning

Choose Your Anchor Wisely: Effective Unlearning Diffusion Models via Concept Reconditioning

Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning

Scaling Laws in Linear Regression: Compute, Parameters, and Data

Transformers as Decision Makers: Provable In-Context Reinforcement Learning via Supervised Pretraining