Zeyi Huang

T2I-ConBench: Text-to-Image Benchmark for Continual Post-training

DynaIP: Dynamic Image Prompt Adapter for Scalable Zero-shot Personalized Text-to-Image Generation

VELR: Efficient Video Reward Feedback via Ensemble Latent Reward Models

MultiCrafter: High-Fidelity Multi-Subject Generation via Spatially Disentangled Attention and Identity-Aware Reinforcement Learning

BideDPO: Conditional Image Generation with Simultaneous Text and Condition Alignment

Manifold Constraint Reduces Exposure Bias in Accelerated Diffusion Sampling