Zhijie Deng
Tenured Track Associate Professor
Works on generative models, including diffusion and efficient large language model decoding used in industrial models such as Meituan LongCat.
Zhijie Deng's lab at SJTU works on generative & foundation models, large language models and image & video generation. 47 papers since 2023, 38 at top venues, mostly ICML and NeurIPS. 6 PhD students and 7 master's students. 5 alumni, most often to ByteDance and OPPO AI Center. Applications close .
At a glance
- papers, 2023–2026
- 12/ year
- at T1 venues
- 81%
- PhD students
- 6
- MS students
- 7
- alumni listed
- 5
Papers by yearfrom the lab homepage, 2023–2026
- 15 papers2026 · T1 14 · collecting
- 14 papers2025 · T1 11
- 12 papers2024 · T1 9
- 6 papers2023 · T1 4
T1 venuesOtherVenues: ICML, NeurIPS, ICLR, CVPR, Findings of EMNLP
Research focuspaper titles that use each topic's words
- Large language models18 papers
- Generative & foundation models11 papers
- Image & video generation9 papers
- decoding4 papers
- used0 papers
Paper list
From the lab homepage.
202615 papers
- EMNLP (T1)CForce: Boosting Parallel Decoding for dLLMs via Consistency Forcing (opens in a new tab)
- ICML (T1)d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillation (opens in a new tab)
- ICLR (T1)Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing (opens in a new tab)
- ICML (T1)Fast and Accurate Causal Parallel Decoding using Jacobi Forcing (opens in a new tab)
- NeurIPS (T1)LatentUM: Unleashing the Potential of Interleaved Cross-Modal Reasoning via a Latent-Space Unified Model (opens in a new tab)
- ICML (T1)LightningRL: Breaking the Accuracy–Parallelism Trade-off of Block-wise dLLMs via Reinforcement Learning (opens in a new tab)
- CVPR (T1)LogCD: Local-to-global Consistency Distillation for Few-step Image Generation (opens in a new tab)
- CVPR (T1)Mantis: A Versatile Vision-Language-Action Model with Disentangled Visual Foresight (opens in a new tab)
- ECCV (T1)MirrorPPR: Exemplar-Based Portrait Photo Retouching (opens in a new tab)
- EMNLP (T1)Multi-Block Diffusion Language Models (opens in a new tab)
- KDD (T1)ProductWebGen: Benchmarking Multimodal Product Webpage Generation (opens in a new tab)
- ICML (T1)Think-Then-Generate: Reasoning-Aware Text-to-Image Diffusion with LLM Encoders (opens in a new tab)
202514 papers
- ICLR (T1)3D-Properties: Identifying Challenges in DPO and Charting a Path Forward (opens in a new tab)
- NeurIPS (T1)Adaptive Discretization for Consistency Models (opens in a new tab)
- ICCV (T1)Advancing Text-to-3D Generation with Linearized Lookahead Variational Score Distillation (opens in a new tab)
- Findings of EMNLPFast Quiet-STaR: Thinking Without Thought Tokens (opens in a new tab)
- SCISLOVECon: Text-driven Training-free Long Video Editing with ControlNet (opens in a new tab)
- ICLR (T1)MatryoshkaKV: Adaptive KV Compression via Trainable Orthogonal Projection (opens in a new tab)
- ICCV (T1)Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts (opens in a new tab)
- TPAMI (T1)Neural Eigenfunctions Are Structured Representation Learners (opens in a new tab)
- ICML (T1)Orthus: Autoregressive Interleaved Image-Text Generation with Modality-Specific Heads (opens in a new tab)
- NeurIPS (T1)Scaling Speculative Decoding with Lookahead Reasoning (opens in a new tab)
- AAAI (T1)SCott: Accelerating Diffusion Models with Stochastic Consistency Distillation (opens in a new tab)
- Findings of EMNLPSIFT: Grounding LLM Reasoning in Contexts via Stickers (opens in a new tab)
202412 papers
- TKDDAccurate and Reliable Forecasting using Stochastic Differential Equations (opens in a new tab)
- Findings of EMNLPAdaMOE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models (opens in a new tab)
- NeurIPS (T1)Amortized Fourier Neural Operators (opens in a new tab)
- ICLR (T1)BayesDiff: Estimating Pixel-wise Uncertainty in Diffusion via Bayesian Inference (opens in a new tab)
- CVPR (T1)Bayesian Exploration of Pre-trained Models for Low-shot Image Classification (opens in a new tab)
- TMLR (T1)Calibrating Deep Ensemble through Functional Variational Inference (opens in a new tab)
- ICML (T1)CLLMs: Consistency Large Language Models (opens in a new tab)
- Findings of ACLEfficient Detection of LLM-generated Texts with a Bayesian Surrogate Model (opens in a new tab)
- ICML (T1)Improved Operator Learning by Orthogonal Attention (opens in a new tab)
- ECCV (T1)Lost in Translation: Latent Concept Misalignment in Text-to-Image Diffusion Models (opens in a new tab)
- ICML (T1)Online Speculative Decoding (opens in a new tab)
- ICML (T1)SpikeZIP-TF: Conversion is All You Need for Transformer-based SNN (opens in a new tab)
20236 papers
- AI OpenBatch Virtual Adversarial Training for Graph Convolutional Networks (opens in a new tab)
- Machine LearningHeterogeneous Multi-Task Gaussian Cox Processes (opens in a new tab)
- ICCV (T1)Learning Neural Eigenfunctions for Unsupervised Semantic Segmentation (opens in a new tab)
- NeurIPS (T1)Learning Sample Difficulty from Pre-trained Models for Reliable Prediction (opens in a new tab)
- NeurIPS (T1)On Calibrating Diffusion Probabilistic Models (opens in a new tab)
- NeurIPS (T1)Towards Accelerated Model Training via Bayesian Data Selection (opens in a new tab)
Where alumni went5 listed, 67% to industry
Industry 2Academia 1
Most often to
- ByteDance
- OPPO AI Center
Also to
- CUHK
Apply to SJTUFall 2027
Computer Science PhD (international applicants)
- Round
- Self-funded
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USMIT
Alexandru Damian
- Learning theory & optimization
- CNZhejiang
Ziqi Yang
- Trustworthy & safe ML
- CNTsinghua
Shu-Tao Xia
- Trustworthy & safe ML
- USCMU
Ruslan Salakhutdinov
- Reinforcement learning
- SGNTU
Tat-Jen Cham
- Generative & foundation models
- USNYU
Gautam Kamath
- Learning theory & optimization
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
SJTUcloses