Stefano Ermon

DiffusionNFT: Online Diffusion Reinforcement with Forward Process

InfoTok: Adaptive Discrete Video Tokenizer via Information-Theoretic Compression

CMT: Mid-Training for Efficient Learning of Consistency, Mean Flow, and Flow-Map Models

Self-Speculative Decoding Accelerates Lossless Inference in Any-Order and Any-Subset Autoregressive Models

Improved Object-Centric Diffusion Learning with Registers and Contrastive Alignment

Principled RL for Diffusion LLMs Emerges from a Sequence-Level Perspective

reAR: Rethinking Visual Autoregressive Models via Token-wise Consistency Regularization

FSPO: Few-Shot Optimization of Synthetic Preferences Effectively Personalizes to Real Users

Align Your Structures: Generating Trajectories with Structure Pretraining for Molecular Dynamics

Inference-Time Scaling of Diffusion Language Models with Particle Gibbs Sampling

RelDiff: Relational Data Generative Modeling with Graph-Based Diffusion Models

DistillKac: Few-Step Image Generation via Damped Wave Equations

Discrete Diffusion Trajectory Alignment via Stepwise Decomposition

GeoEvolve: Automating Geospatial Model Discovery via Multi-Agent Large Language Models

RFG: Test-Time Scaling for Diffusion Large Language Model Reasoning with Reward-Free Guidance

Divergence Minimization Preference Optimization for Diffusion Model Alignment

Improving Diffusion Language Model Reasoning through Joint Search in Generation Order and Token Space

Inductive Moment Matching

GeoAda: Efficiently Finetune Geometric Diffusion Models with Equivariant Adapters

Exploring Diffusion Transformer Designs via Grafting

Training-Free Safe Denoisers for Safe Use of Diffusion Models

Energy-Based Diffusion Language Models for Text Generation

Preference-Guided Diffusion for Multi-Objective Offline Optimization

Sample Efficient Preference Alignment in LLMs via Active Exploration

TFG-Flow: Training-free Guidance in Multimodal Generative Flow

ExPLoRA: Parameter-Efficient Extended Pre-Training to Adapt Vision Transformers under Domain Shifts

Smooth Interpolation for Improved Discrete Graph Generative Models

TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation

Data Unlearning in Diffusion Models

Consistency Flow Matching: Defining Straight Flows with Velocity Consistency

Scaling Probabilistic Circuits via Monarch Matrices

ExPLoRA: Parameter-Efficient Extended Pre-Training to Adapt Vision Transformers under Domain Shifts

G2D2: Gradient-guided Discrete Diffusion for image inverse problem solving

Zero-Shot Cyclic Peptide Design via Composable Geometric Constraints

Sharpe Ratio-Guided Active Learning for Preference Optimization in RLHF

Smooth Probabilistic Interpolation Benefits Generative Modeling for Discrete Graphs

SSIF: Physics-Inspired Implicit Representations for Spatial-Spectral Image Super-Resolution

Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation

Generalizing Stochastic Smoothing for Differentiation and Gradient Estimation

TEOChat: A Large Vision-Language Assistant for Temporal Earth Observation Data

Efficient Scaling of Diffusion Transformers for Text-to-Image Generation

Bellman Diffusion: Generative Modeling as Learning a Linear Operator in the Distribution Space

CPSample: Classifier Protected Sampling for Guarding Training Data During Diffusion

Nonmyopic Bayesian Optimization in Dynamic Cost Settings

Trans4D: Realistic Geometry-Aware Transition for Compositional Text-to-4D Synthesis

Mitigating Embedding Collapse in Diffusion Models for Categorical Data