weiminbai@stu.pku.edu.cn Greetings! I am a PhD Candidate majoring in Computer Science and Physics at Peking University, supervised by Dr. He Sun and Dr. Wenzheng Chen. I specialize in generative models and inverse rendering. I developed EMDiffusion for data-efficient diffusion training, LatentDEM for blind inversion, VLM3D and Dive3D for 3D-aware reward modeling, and InstantViR for real-time video reconstruction. I have held internships and collaborations with leading industrial institutions such as Microsoft Research Asia, RedNote Inc., Kuaishou Inc., and Shanghai AI Laboratory. I obtained my Master's degree in Computer Science from Southeast University in 2023, where I was advised by Prof. Hui Xue. Prior to that, I earned a Bachelor's degree in Computer Science from China Agricultural University in 2020. I am actively seeking internship for 2026 and full-time positions starting in Summer 2027. I am open to collaborations and discussions—please feel free to reach out anytime. Vision Build practical intelligence, which necessitates (1) real-time inference and (2) infinite context memory, thus ultimately empowering applications like agents to liberate human society from repetitive labor and enjoy a peaceful life. News May, 2026: VLM3D gets accepted to ICML 2026. (Bai et al., 2026) Apr, 2026: InstantViR is selected as a CVPR 2026 Highlight, and I receive a CVPR 2026 Scholarship. Feb, 2026: Two papers get accepted to CVPR 2026. (Bai et al., 2026) (Gu et al., 2026) Dec, 2025: One paper gets accepted to Transactions on Machine Learning Research (TMLR). (Bai et al., 2025) Dec, 2025: One paper gets accepted to IEEE TIP. (Bai et al., 2025) Nov, 2025: One poster gets accepted to SIGGRAPH ASIA 2025. (Bai et al., 2025) Sep, 2025: One paper gets accepted to NeurIPS 2025. (Wang et al., 2025) Apr, 2025: Oral presentation at ICASSP 2025, India. (Bai et al., 2025) Jan, 2025: Long Oral presentation at SPIE Photonics West 2025, San Francisco. (Bai et al., 2025) Sep, 2024: One paper gets accepted to NeurIPS 2024. (Bai et al., 2024) Show more 2025: EM Generalist: A physics-driven diffusion foundation model for electron microscopy. H Sun, E Ye, C Lu, Z Jiang, W Bai, S Ren, C Wang, J Zhang, R Shi, L Ma, ... (2025) 2025: Microscopic image restoration and uncertainty quantification using physics-informed generative models. W Bai, S Ren, E Ye, W Tang, H Sun. Computational Optical Imaging and Artificial Intelligence in Biomedical …, 2025 2025: InstantViR: Real-Time Video Inverse Problem Solver with Distilled Diffusion Prior. W Bai, S Xu, Y Ren, J Hao, M Sun, W Chen, H Sun. arXiv preprint arXiv:2511.14208, 2025 2025: Masked Auto-Regressive Variational Acceleration: Fast Inference Makes Practical Reinforcement Learning. Y Gu, W Bai, Y Wang, W Luo, H Sun. arXiv preprint arXiv:2511.15190, 2025 2025: Let Language Constrain Geometry: Vision-Language Models as Semantic and Spatial Critics for 3D Generation. W Bai, Y Li, W Luo, Z Lai, Y Wang, W Chen, H Sun. arXiv preprint arXiv:2511.14271, 2025 August 2024: Talk at University of Chicago IMSI — Solving Inverse Problems with Inaccurate Priors or Physical Model. 2025: Dive3D: Diverse Distillation-based Text-to-3D Generation via Score Implicit Matching. W Bai, Y Li, W Chen, W Luo, H Sun. Transactions on Machine Learning Research, 2025 2025: Vision-Language Models as Differentiable Semantic and Spatial Rewards for Text-to-3D Generation. W Bai, Y Li, W Luo, W Chen, H Sun. arXiv preprint arXiv:2509.15772, 2025 2025: Blind inversion using latent diffusion priors. W Bai, S Chen, W Chen, H Sun. SIGGRAPH Asia 2025 Poster (IEEE Transactions on Image Processing), 2025 2025: Uni-Instruct: One-step Diffusion Model through Unified Diffusion Divergence Instruction. Y Wang, W Bai, C Zhang, D Zhang, W Luo, H Sun. Advances in Neural Information Processing Systems, 2025 2025: Learning diffusion model from noisy measurement using principled expectation-maximization method. W Bai, W Tang, E Ye, S Chen, W Chen, H Sun. ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and …, 2025 2024: An expectation-maximization algorithm for training clean diffusion models from corrupted observations. W Bai, Y Wang, W Chen, H Sun. Advances in Neural Information Processing Systems, 2024 2024: Integrating amortized inference with diffusion models for learning clean distribution from corrupted images. Y Wang, W Bai, W Luo, W Chen, H Sun. arXiv preprint arXiv:2407.11162, 2024 2023: A temporal kernel for time series forecasting. W Bai, H Xue. Knowledge-Based Systems, 113623, 2023 2021: Learning Deeper Non-Monotonic Networks by Softly Transferring Solution Space. ZF Wu, H Xue, W Bai. IJCAI, 3200-3206, 2021 Selected Publications (full list see Google Scholar) ICML 2026 Let Language Constrain Geometry: Vision-Language Models as Semantic and Spatial Critics for 3D Generation Weimin Bai, Yubo Li, Weijian Luo, Zeqiang Lai, Yequan Wang, Wenzheng Chen, He Sun ICML 2026 [Page] [Paper] [Code] CVPR 2026 InstantViR: Real-Time Video Inverse Problem Solver with Distilled Diffusion Prior Weimin Bai, Suzhe Xu, Yiwei Ren, Jinhua Hao, Ming Sun, Wenzheng Chen, He Sun CVPR 2026 Highlight A real-time video super-resolution, inpainting, and deblurring framework utilizing amortized autoregressive distillation for fast inference. [Page] [Paper] [Code] CVPR 2026 Masked Auto-Regressive Variational Acceleration: Fast Inference Makes Practical Reinforcement Learning Yuxuan Gu, Weimin Bai, Yifei Wang, Weijian Luo, He Sun CVPR 2026 The first reinforcement learning framework for masked auto-regressive models. [Paper] TMLR 2025 Dive3D: Diverse Distillation-based Text-to-3D Generation via Score Implicit Matching Weimin Bai, Yubo Li, Weijian Luo, Wenzheng Chen, He Sun Transactions on Machine Learning Research (TMLR), 2025 [Page] [Paper] [Code] IEEE TIP 2025 Blind Inversion using Latent Diffusion Priors Weimin Bai, Siyi Chen, Wenzheng Chen, He Sun IEEE Transactions on Image Processing, 2025 (SIGGRAPH Asia 2025 Poster) [Page] [Paper] [Code] NeurIPS 2025 Uni-Instruct: One-step Diffusion Model through Unified Diffusion Divergence Instruction Yifei Wang, Weimin Bai, Colin Zhang, Debing Zhang, Weijian Luo, He Sun NeurIPS, 2025 [Paper] [Code] SPIE Photonics West 2025 Microscopic Image Restoration and Uncertainty Quantification using Physics-informed Generative Models Weimin Bai, Yubo Li, Wenzheng Chen, Weijian Luo, He Sun SPIE Photonics West (Long Oral Presentation), 2025 [Page] NeurIPS 2024 An Expectation-Maximization Algorithm for Training Clean Diffusion Models from Corrupted Observations Weimin Bai, Yifei Wang, Wenzheng Chen, He Sun NeurIPS, 2024 [Paper] [Code] KBS 2023 A Temporal Kernel for Time Series Forecasting Weimin Bai, Hui Xue Knowledge-Based Systems, 2023 Mentors with whom I have worked on projects: He Sun, PhD, tenure-track Assistant Professor at Peking University. Wenzheng Chen, PhD, tenure-track Assistant Professor at Peking University. Weijian Luo, PhD, senior research scientist at hi-lab, RedNote, aka Xiaohongshu Inc. Hui Xue, Professor of Computer Science Department, Southeast University, Vice Dean and Director of PALM Lab. Debing Zhang, PhD, Director of Artificial General Intelligence (AGI) team of RedNote, aka Xiaohongshu Inc. Friends with whom I have worked with: Yifei Wang, Undergraduate student at Peking University, and current PhD student at Rice University (Houston). Siyi Chen, Undergraduate student at Peking University, and current PhD student at University of Michigan (Ann Arbor). Weiheng Tang, Undergraduate student at Peking University, and current PhD student at Purdue University. Yuxuan Gu, Undergraduate student at Beihang University, M.S. student at Peking University. Yubo Li, Undergraduate student at Tsinghua University, M.S. student at Peking University. Suzhe Xu, Undergraduate student at Huaqiao University, current internship at Kuaishou Inc. Academic Service Journal Reviewer: IEEE Transactions on Image Processing (TIP), IEEE Transactions on Neural Networks and Learning Systems (TNNLS), Knowledge-Based Systems (KBS), Conference Reviewer: NeurIPS 2025, ICLR 2026/2025, ICASSP 2024/2025/2026, IJCAI 2023.