Zhenguo Li

Variational Autoencoding Discrete Diffusion with Enhanced Dimensional Correlations Modeling

Mathesis: Towards Formal Theorem Proving from Natural Languages

Reasoning-Aligned Perception Decoupling for Scalable Multi-modal Reasoning

Any-Order GPT as Masked Diffusion Model: Decoupling Formulation and Architecture

Agreement with the Ensemble for Zero-Shot Vision-Language Model Selection

DreamOn: Diffusion Language Models For Code Infilling Beyond Fixed-size Canvas

Masked Diffusion Models as Energy Minimization

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

CARTS: Advancing Neural Theorem Proving with Diversified Tactic Calibration and Bias-Resistant Tree Search

Multi-Objective One-Shot Pruning for Large Language Models

FUDOKI: Discrete Flow-based Unified Understanding and Generation via Kinetic-Optimal Velocities

ProofAug: Efficient Neural Theorem Proving via Fine-grained Proof Structure Analysis

Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Data

Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planning

The Buffer Mechanism for Multi-Step Information Reasoning in Language Models

Improved Diffusion-based Generative Model with Better Adversarial Robustness

Forewarned is Forearmed: Harnessing LLMs for Data Synthesis via Failure-induced Exploration

Jailbreaking as a Reward Misspecification Problem

Accelerating Auto-regressive Text-to-Image Generation with Training-free Speculative Jacobi Decoding

DAPE V2: Process Attention Score as Feature Map for Length Extrapolation

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

G-LLaVA: Solving Geometric Problem with Multi-Modal Large Language Model

EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions

MagicDrive3D: Controllable 3D Generation for Any-View Rendering in Street Scenes

Implicit Search via Discrete Diffusion: A Study on Chess

Mixture of insighTful Experts (MoTE): The Synergy of Thought Chains and Expert Mixtures in Safety Self-Alignment

Efficient Multi-modal Large Language Models via Visual Token Grouping

Recovering Knowledge by Hardening Language Models