Fu-Yun Wang

PhD student@The Chinese University of Hong Kong·中国香港·OpenReview

研究方向

Reinforcement Learning; Unified Multimodal Models; Agentic Learning · Diffusion Models; Consistency Models; Rectified Flow; Video & Image Synthesis; · Incremental Learning; Continual Learning

7.3

Speculative Jacobi-Denoising Decoding for Accelerating Autoregressive Text-to-image Generation

NeurIPS 2025Poster

二作

7.0

Diffusion-NPO: Negative Preference Optimization for Better Preference Aligned Generation of Diffusion Models

ICLR 2025Poster

一作

6.7

InstantPortrait: One-Step Portrait Editing via Diffusion Multi-Objective Distillation

ICLR 2025Poster

三作

5.8

Rectified Diffusion: Straightness Is Not Your Need in Rectified Flow

ICLR 2025Poster

一作

4.0

Trans4D: Realistic Geometry-Aware Transition for Compositional Text-to-4D Synthesis

合作者 (20)

Fu-Yun Wang

Speculative Jacobi-Denoising Decoding for Accelerating Autoregressive Text-to-image Generation

Diffusion-NPO: Negative Preference Optimization for Better Preference Aligned Generation of Diffusion Models

InstantPortrait: One-Step Portrait Editing via Diffusion Multi-Objective Distillation

Rectified Diffusion: Straightness Is Not Your Need in Rectified Flow

Trans4D: Realistic Geometry-Aware Transition for Compositional Text-to-4D Synthesis

Stable Consistency Tuning: Understanding and Improving Consistency Models

Phased Consistency Models

Lumina-Next : Making Lumina-T2X Stronger and Faster with Next-DiT