Zhijie Deng

Associate Professor@Shanghai Jiaotong University·中国·OpenReview

研究方向

AI for PDEs · Large language models · Diffusion models · spectral learning · kernel approximation · Deep generative model · Bayesian neural networks · Neural architecture search

Zhijie Deng

Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing

Reinforcing Query-Level Meta-Agents

Done Is Better than Perfect: Unlocking Efficient Reasoning by Structured Multi-Turn Decomposition

MMWebGen: Benchmarking Multimodal Webpage Generation

Exploring Aleatoric Uncertainty in Object Detection via Vision Foundation Models

UniCMs: A Unified Consistency Model For Efficient Multimodal Generation and Understanding

AR Models can be Faster and More Accurate Parallel Decoders than Diffusion LLMs

Which Data Attributes Stimulate Math and Code Reasoning? An Investigation via Influence Functions

Adaptive Discretization for Consistency Models

3D-Properties: Identifying Challenges in DPO and Charting a Path Forward

Orthus: Autoregressive Interleaved Image-Text Generation with Modality-Specific Heads

MatryoshkaKV: Adaptive KV Compression via Trainable Orthogonal Projection

Scaling Speculative Decoding with Lookahead Reasoning

Efficient-vDiT: Efficient Video Diffusion Transformers With Attention Tile

On Exploring Visual Attention Shrinking for Accelerating VLMs for Video Understanding

Advancing Text-to-3D Generation with Linearized Lookahead Variational Score Distillation

TLCM: Training- efficient Latent Consistency Model for Image Generation with 2-8 Steps

In-context KV-Cache Eviction for LLMs via Attention-Gate

Guided-BFNs: Towards Visualizing and Understanding Bayesian Flow Networks in the Context of Trajectory Planning

Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts