Yinchuan Li

EmbodiedMAE: A Unified 3D Multi-Modal Representation for Robot Manipulation

DiMeR: Disentangled Mesh Reconstruction Model with Normal-only Geometry Training

PhysToolBench: Benchmarking Physical Tool Understanding for MLLMs

No Place to Hide: Benchmarking Video Hallucination with Background-Controlled Pairs

Proximalized Preference Optimization for Diverse Feedback Types: A Decomposed Perspective on DPO

Conditioning Matters: Training Diffusion Policies is Faster Than You Think

STAR: Learning Diverse Robot Skill Abstractions through Rotation-Augmented Vector Quantization

Two-Steps Diffusion Policy for Robotic Manipulation via Genetic Denoising

Ergodic Generative Flows

3D-AffordanceLLM: Harnessing Large Language Models for Open-Vocabulary Affordance Detection in 3D Worlds

A Theory of Multi-Agent Generative Flow Networks