publications

2026

1. ICLR

   

   Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency

   Kaiwen Zheng , Yuji Wang, Qianli Ma , Huayu Chen , Jintao Zhang, Yogesh Balaji, Jianfei Chen, Ming-Yu Liu^†, Jun Zhu^†, and Qinsheng Zhang

   In The Fourteenth International Conference on Learning Representations, 2026

   PDF Code Website
2. ICLR

   

   SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse-Linear Attention

   Jintao Zhang , Haoxu Wang, Kai Jiang, Shuo Yang, Kaiwen Zheng, Haocheng Xi , Ziteng Wang , Hongzhou Zhu, Min Zhao, Ion Stoica, Joseph E. Gonzalez, Jun Zhu^†, and Jianfei Chen^†

   In The Fourteenth International Conference on Learning Representations, 2026

   PDF Code
3. ICLR

   

   DiffusionNFT: Online Diffusion Reinforcement with Forward Process

   Kaiwen Zheng^* , Huayu Chen^*, Haotian Ye , Haoxiang Wang, Qinsheng Zhang, Kai Jiang, Hang Su, Stefano Ermon, Jun Zhu^†, and Ming-Yu Liu^†

   In The Fourteenth International Conference on Learning Representations, 2026

   Oral, Top 1.2%

   PDF Code Website
4. ICLR

   

   Bridging Supervised Learning and Reinforcement Learning in Math Reasoning

   Huayu Chen, Kaiwen Zheng, Qinsheng Zhang, Ganqu Cui, Yin Cui, Haotian Ye, Tsung-Yi Lin, Ming-Yu Liu^†, Jun Zhu^† , and Haoxiang Wang

   In The Fourteenth International Conference on Learning Representations, 2026

   PDF Code Website

2025

1. arXiv

   

   TurboDiffusion: Accelerating Video Diffusion Models by 100-200 Times

   Jintao Zhang^*, Kaiwen Zheng^*, Kai Jiang^* , Haoxu Wang^*, Ion Stoica, Joseph E Gonzalez, Jianfei Chen, and Jun Zhu^†

   Technical Report, 2025

   PDF Code
2. arXiv

   

   World Simulation with Video Foundation Models for Physical AI

   NVIDIA

   Technical Report, 2025

   PDF Code Website
3. ICML

   

   Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator

   Kaiwen Zheng , Yongxin Chen , Huayu Chen, Guande He, Ming-Yu Liu^†, Jun Zhu^†, and Qinsheng Zhang

   In Proceedings of the 42nd International Conference on Machine Learning, 2025

   Spotlight, Top 2.6%

   PDF Code Website

   

   
4. ICML

   

   Visual Generation Without Guidance

   Huayu Chen^*, Kai Jiang^*, Kaiwen Zheng, Jianfei Chen, Hang Su, and Jun Zhu^†

   In Proceedings of the 42nd International Conference on Machine Learning, 2025

   PDF Code
5. ICLR

   

   Elucidating the Preconditioning in Consistency Distillation

   Kaiwen Zheng^*, Guande He^*, Jianfei Chen, Fan Bao, and Jun Zhu^†

   In The Thirteenth International Conference on Learning Representations, 2025

   PDF
6. ICLR

   

   Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling

   Kaiwen Zheng , Yongxin Chen, Hanzi Mao, Ming-Yu Liu^†, Jun Zhu^†, and Qinsheng Zhang

   In The Thirteenth International Conference on Learning Representations, 2025

   Top 5%

   PDF Blog
7. ICLR

   

   Diffusion Bridge Implicit Models

   Kaiwen Zheng^*, Guande He^*, Jianfei Chen, Fan Bao, and Jun Zhu^†

   In The Thirteenth International Conference on Learning Representations, 2025

   PDF Code

2024

1. NeurIPS

   

   Consistency Diffusion Bridge Models

   Guande He^*, Kaiwen Zheng^*, Jianfei Chen, Fan Bao, and Jun Zhu^†

   In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

   PDF Code
2. NeurIPS

   

   Aligning Diffusion Behaviors with Q-functions for Efficient Continuous Control

   Huayu Chen, Kaiwen Zheng, Hang Su, and Jun Zhu^†

   In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

   PDF Code
3. NeurIPS

   

   Identifying and Solving Conditional Image Leakage in Image-to-Video Diffusion Model

   Min Zhao^* , Hongzhou Zhu^*, Chendong Xiang, Kaiwen Zheng, Chongxuan Li^†, and Jun Zhu^†

   In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

   PDF Code Website
4. arXiv

   

   Vidu: a Highly Consistent, Dynamic and Skilled Text-to-Video Generator with Diffusion Models

   Fan Bao, Chendong Xiang^*, Gang Yue^*, Guande He^* , Hongzhou Zhu^*, Kaiwen Zheng^*, Min Zhao^* , Shilong Liu^* , Yaole Wang^*, and Jun Zhu^†

   Technical Report, 2024

   PDF Website
5. ICLR

   

   InstructPix2NeRF: Instructed 3D Portrait Editing from a Single Image

   Jianhui Li , Shilong Liu , Zidong Liu , Yikai Wang, Kaiwen Zheng, Jinghui Xu , Jianmin Li^†, and Jun Zhu^†

   In The Twelfth International Conference on Learning Representations, 2024

   PDF Code Website

2023

1. arXiv

   

   Schrodinger Bridges Beat Diffusion Models on Text-to-Speech Synthesis

   Zehua Chen^*, Guande He^*, Kaiwen Zheng^*, Xu Tan, and Jun Zhu^†

   arXiv preprint arXiv:2312.03491, 2023

   PDF Code Website
2. NeurIPS

   

   DPM-Solver-v3: Improved Diffusion ODE Solver with Empirical Model Statistics

   Kaiwen Zheng^*, Cheng Lu^*, Jianfei Chen, and Jun Zhu^†

   In Advances in Neural Information Processing Systems, 2023

   PDF Code Slides Website
3. ICML

   

   Improved Techniques for Maximum Likelihood Estimation for Diffusion ODEs

   Kaiwen Zheng^*, Cheng Lu^*, Jianfei Chen, and Jun Zhu^†

   In Proceedings of the 40th International Conference on Machine Learning, 2023

   PDF Code

   
4. CVPR

   

   PREIM3D: 3d Consistent Precise Image Attribute Editing From a Single Image

   Jianhui Li , Jianmin Li^† , Haoji Zhang , Shilong Liu, Zhengyi Wang, Zihao Xiao, Kaiwen Zheng, and Jun Zhu^†

   In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023

   PDF Code Website

2022

1. ICML

   

   Maximum Likelihood Training for Score-Based Diffusion ODEs by High Order Denoising Score Matching

   Cheng Lu, Kaiwen Zheng, Fan Bao, Jianfei Chen^†, Chongxuan Li, and Jun Zhu^†

   In Proceedings of the 39th International Conference on Machine Learning, 2022

   PDF Code