Rui Pan

GAR: Generative Adversarial Reinforcement Learning for Formal Theorem Proving

Lean4Physics: Comprehensive Reasoning Framework for College-level Physics in Lean4

Unbiased Gradient Low-Rank Projection

Generalizable Geometric Image Caption Synthesis

ERA: Transforming VLMs into Embodied Agents via Embodied Prior Learning and Online Reinforcement Learning

AdaGrad under Anisotropic Smoothness

Personalized Visual Instruction Tuning

ASGO: Adaptive Structured Gradient Optimization

Understanding Overadaptation in Supervised Fine-Tuning: The Role of Ensemble Methods

MA-LoT: Model-Collaboration Lean-based Long Chain-of-Thought Reasoning enhances Formal Theorem Proving