Haoran Li

6 papers University of the Chinese Academy of Sciences University of Illinois Urbana-Champaign

• HoPE: Hybrid of Position Embedding for Long Context Vision-Language Models
• Latent Mixture of Symmetries for Sample-Efficient Dynamic Learning
• Purity Law for Neural Routing Problem Solvers with Enhanced Generalizability
• TANDEM: Bi-Level Data Mixture Optimization with Twin Networks
• The Primacy of Magnitude in Low-Rank Adaptation
• Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations