mlsanityData to 14 Oct 2026

CREPE: Controlling Diffusion with Replica Exchange

arXiv 2509.23265 ↗First posted 27 Sep 2025cs.LGcited by 11

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Inference-time control of diffusion models aims to steer model outputs to satisfy new constraints without retraining. Previous approaches have mostly relied on heuristic guidance or have been coupled with Sequential Monte Carlo (SMC) for bias correction. In this paper, we propose a flexible alternative based on replica exchange, an algorithm designed initially for sampling problems. We refer to this method as CREPE (Controlling with REPlica Exchange). Unlike SMC, CREPE: (1) generates particles sequentially, (2) maintains high diversity in the generated samples after a burn-in period, and (3) enables online refinement or early termination. We demonstrate its versatility across various tasks, including temperature annealing, reward-tilting, model composition and classifier-free guidance debiasing, with competitive performance compared to prior SMC methods.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 48 of its 81 references that are in the map. First posted on arXiv 27 Sep 2025.

Builds on

48 of 81 references.

Closest ancestors · 21

Same direction, newest first

  1. 2025Progressive Inference-Time Annealing of Diffusion Models for Sampling from Boltzmann Densities19 Jun
  2. 2025Progressive Tempering Sampler with Diffusion5 Jun
  3. 2025Feynman-Kac Correctors in Diffusion: Annealing, Guidance, and Product of Experts4 Mar
  4. 2025Composition and Control with Distilled Energy Diffusion Models and Sequential Monte Carlo18 Feb
  5. 2025Debiasing Guidance for Discrete Diffusion with Sequential Monte Carlo10 Feb
  6. 2025A General Framework for Inference-time Scaling and Steering of Diffusion Models12 Jan
  7. 2024TFG: Unified Training-Free Guidance for Diffusion Models24 Sep
  8. 2024Derivative-Free Guidance in Continuous and Discrete Diffusion Models with Soft Value-Based Decoding15 Aug
  9. 2024DEFT: Efficient Fine-Tuning of Diffusion Models by Learning the Generalised h-transform3 Jun
  10. 2024Diffusion Models as Constrained Samplers for Optimization with Unknown Constraints28 Feb
  11. 2023Analyzing and Improving the Training Dynamics of Diffusion Models5 Dec
  12. 2023Reduce, Reuse, Recycle: Compositional Generation with Energy-Based Diffusion Models and MCMC22 Feb
9 more
  1. 2022Classifier-Free Diffusion Guidance26 Jul
  2. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  3. 2022A Continuous Time Framework for Discrete Denoising Models30 May
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  5. 2021Structured Denoising Diffusion Models in Discrete State-Spaces7 Jul
  6. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  7. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
  8. 2020Denoising Diffusion Implicit Models6 Oct
  9. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 27

  1. 2021CLIPScore: A Reference-free Evaluation Metric for Image Captioningcited by 2,424
  2. 2022Video Diffusion Modelscited by 2,305
  3. 2022Diffusion Posterior Sampling for General Noisy Inverse Problemscited by 1,416
  4. 2023ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generationcited by 1,045
  5. 2023Stochastic Interpolants: A Unifying Framework for Flows and Diffusionscited by 939
  6. 2022Equivariant Diffusion for Molecule Generation in 3Dcited by 677
  7. 2024Simplified and Generalized Masked Diffusion for Discrete Datacited by 626
  8. 2024OGBench: Benchmarking Offline Goal-Conditioned RLcited by 238
  9. 2024Adjoint Matching: Fine-tuning Flow and Diffusion Generative Models with Memoryless Stochastic Optimal Controlcited by 233
  10. 2023Practical and Asymptotically Exact Conditional Sampling in Diffusion Modelscited by 232
  11. 2022Structure-based Drug Design with Equivariant Diffusion Modelscited by 217
  12. 2024Iterated Denoising Energy Matching for Sampling from Boltzmann Densitiescited by 116
15 more
  1. 2024Beyond ELBOs: A Large-Scale Evaluation of Variational Methods for Samplingcited by 56
  2. 2023OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentialscited by 47
  3. 2023Learning Interpolations between Boltzmann Densitiescited by 43
  4. 2025Generative Trajectory Stitching through Diffusion Compositioncited by 41
  5. 2024Efficient Diversity-Preserving Diffusion Alignment via Gradient-Informed GFlowNetscited by 32
  6. 2025LEAPS: A discrete neural sampler via locally equivariant networkscited by 31
  7. 2023Accurate transition state generation with an object-aware equivariant elementary reaction diffusion modelcited by 30
  8. 2024Improving GFlowNets for Text-to-Image Diffusion Alignmentcited by 27
  9. 2024Diffusive Gibbs Samplingcited by 21
  10. 2024Training Neural Samplers with Reverse Diffusive KL Divergencecited by 21
  11. 2025From discrete-time policies to continuous-time diffusion samplers: Asymptotic equivalences and faster trainingcited by 19
  12. 2025Accelerated Parallel Tempering via Neural Transportscited by 17
  13. 2025FEAT: Free energy Estimators with Adaptive Transportcited by 16
  14. 2021q-Paths: Generalizing the Geometric Annealing Path using Power Meanscited by 12
  15. 2024On diffusion posterior sampling via sequential Monte Carlo for zero-shot scaffolding of protein motifscited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 11.

11 papers.

  1. 2026SURGE: Approximation and Training Free Particle Filter for Diffusion Surrogatecited by 3
  2. 2026Towards Diverse Scientific Hypothesis Search with Large Language Modelscited by 3
  3. 2026Diffusion Alignment Beyond KL: Variance Minimisation as Effective Policy Optimisercited by 1
  4. 2026Composing diffusion priors with explicit physical context via generative Gibbs samplingcited by 1
  5. 2026Recipes for Steering and Scaling LLMs via Samplingcited by 1
  6. 2026METALICA: METAdynamics and repLICA exchange for enhanced diffusion samplingcited by 1
  7. 2026Bayesian Rain Field Reconstruction using Commercial Microwave Links and Diffusion Model Priorscited by 0
  8. 2026Parallel Tempering Initial Sampling in Inference-Time Reward Alignmentcited by 0
  9. 2026Parallel Tempering for Diffusion-Based Combinatorial Optimizationcited by 0
  10. 2026Traversing the Satisfaction-Diversity Frontier in Text-to-Image Diffusioncited by 0
  11. 2026ENCORE: Exact Non-equilibrium COntrol with Replica Exchange for Diffusion Generationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026A Mean-Field Framework for Inference-Time Distributional Control of Diffusion Models
  2. 2025RNE: plug-and-play diffusion inference-time control and energy-based training
  3. 2025Inference-Time Alignment in Diffusion Models with Reward-Guided Generation: Tutorial and Review
  4. 2026Diversify Diffusion with Temperature Sampling and Variance-Corrective Time Shifting
  5. 2023Transport meets Variational Inference: Controlled Monte Carlo Diffusions
  6. 2025DriftLite: Lightweight Drift Control for Inference-Time Scaling of Diffusion Models
  7. 2025Test-time Alignment of Diffusion Models without Reward Over-optimization
  8. 2026Inference-Time Attribute Distribution Alignment for Unconditional Diffusion