Fangzhen Lin

Reverse-Engineered Reasoning for Open-Ended Generation

Emergent Hierarchical Reasoning in LLMs through Reinforcement Learning

OOTSM: A Decoupled Linguistic Framework for Effective Scene Graph Anticipation

VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning

Pixel Reasoner: Incentivizing Pixel Space Reasoning via Curiosity-Driven Reinforcement Learning

Multi-Task Vehicle Routing Solver via Mixture of Specialized Experts under State-Decomposable MDP

AutoCode4Math: Learning Autonomous Code Integration for Math LLMs