Yuhang Zang

Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Models

STAR-Bench: Probing Deep Spatio-Temporal Reasoning as Audio 4D Intelligence

SIM-CoT: Supervised Implicit Chain-of-Thought

Visual Self-Refine: A Pixel-Guided Paradigm for Accurate Chart Parsing

DiCache: Let Diffusion Model Determine Its Own Cache

Advancing Complex Video Object Segmentation via Progressive Concept Construction

ScaleCap: Scalable Image Captioning via Dual-Modality Debiasing

CODA: Coordinating the Cerebrum and Cerebellum for a Dual-Brain Computer Use Agent with Decoupled Reinforcement Learning

CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning

Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning

Unified Reward Model for Multimodal Understanding and Generation

SPARK: Synergistic Policy And Reward Co-Evolving Framework

SEAgent: Self-Evolving Computer Use Agent with Autonomous Learning from Experience

Demo-ICL: In-Context Learning for Procedural Video Knowledge Acquisition

$\text{G}^2$RPO: Granular GRPO for Precise Reward in Flow Models

Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning

BoostStep: Boosting Mathematical Capability of Large Language Models via Step-aligned In Context Learning

Generative Photographic Control for Scene-Consistent Video Cinematic Editing

VideoRoPE: What Makes for Good Video Rotary Position Embedding?

HiFlow: Training-free High-Resolution Image Generation with Flow-Aligned Guidance

Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning

MotionClone: Training-Free Motion Cloning for Controllable Video Generation

MIA-DPO: Multi-Image Augmented Direct Preference Optimization For Large Vision-Language Models

RAR: Retrieving And Ranking Augmented MLLMs for Visual Recognition

SongGen: A Single Stage Auto-regressive Transformer for Text-to-Song Generation

Deciphering Cross-Modal Alignment in Large Vision-Language Models with Modality Integration Rate

SAM2Long: Enhancing SAM2 for Long Video Segmentation with a Training-Free Memory Tree

BroadWay: Boost Your Text-to-Video Generation Model in a Training-free Way

Bootstrap3D: Improving Multi-view Diffusion Model with Synthetic Data

PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction