mlsanityData to 7 Oct 2026

Inference-time Scaling of Diffusion Models through Classical Search

arXiv 2505.23614 ↗First posted 29 May 2025cs.LG stat.MLcited by 48

On the map: LLM reasoning → Diffusion reasoning and test-time scaling

also in 1 more
Abstract (arXiv)

Classical search algorithms have long underpinned modern artificial intelligence. In this work, we tackle the challenge of inference-time control in diffusion models -- adapting generated outputs to meet diverse test-time objectives -- using principles from classical search. We propose a general framework that orchestrates local and global search to efficiently navigate the generative space. It employs a theoretically grounded local search via annealed Langevin MCMC and performs compute-efficient global exploration using breadth-first and depth-first tree search. We evaluate our approach on a range of challenging domains, including planning, offline reinforcement learning, and image generation. Across all tasks, we observe significant gains in both performance and efficiency. These results show that classical search provides a principled and practical foundation for inference-time scaling in diffusion models. Project page at https://diffusion-inference-scaling.github.io/.

Key facts. Cited by 48 papers in the map (data through 7 Oct 2026). Placed in Diffusion reasoning and test-time scaling, one of 17 directions in LLM reasoning. Builds on 87 of its 110 references that are in the map. First posted on arXiv 29 May 2025.

Builds on

87 of 110 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025Scaling Image and Video Generation via Test-Time Evolutionary Search23 May
  2. 2025Adaptive Inference-Time Scaling via Cyclic Diffusion Search20 May
  3. 2025Reflect-DiT: Inference-Time Scaling for Text-to-Image Diffusion Transformers via In-Context Reflection15 Mar
  4. 2025Monte Carlo Tree Diffusion for System 2 Planning11 Feb
  5. 2025SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer30 Jan
  6. 2025Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps16 Jan
  7. 2025A General Framework for Inference-time Scaling and Steering of Diffusion Models12 Jan

Other references · 80

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  3. 2020Denoising Diffusion Implicit Modelscited by 9,655
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  5. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  6. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  7. 2022Classifier-Free Diffusion Guidancecited by 5,717
  8. 2023Adding Conditional Control to Text-to-Image Diffusion Modelscited by 5,466
  9. 2022Flow Matching for Generative Modelingcited by 5,426
  10. 2013Intriguing properties of neural networkscited by 5,114
  11. 2020Training data-efficient image transformers & distillation through attentioncited by 4,560
  12. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,310
38 more
  1. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,290
  2. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,094
  3. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  4. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,736
  5. 2023Let's Verify Step by Stepcited by 3,598
  6. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  7. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,730
  8. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,362
  9. 2022Do As I Can, Not As I Say: Grounding Language in Robotic Affordancescited by 2,302
  10. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  11. 2022Video Diffusion Modelscited by 2,280
  12. 2022DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Stepscited by 1,961
  13. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,928
  14. 2020Conservative Q-Learning for Offline Reinforcement Learningcited by 1,777
  15. 2015Deep Reinforcement Learning with Double Q-learningcited by 1,704
  16. 2024Octo: An Open-Source Generalist Robot Policycited by 1,427
  17. 2020D4RL: Datasets for Deep Data-Driven Reinforcement Learningcited by 1,406
  18. 2021Offline Reinforcement Learning with Implicit Q-Learningcited by 1,245
  19. 2018Off-Policy Deep Reinforcement Learning without Explorationcited by 1,227
  20. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,023
  21. 2023ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generationcited by 1,010
  22. 2022Building Normalizing Flows with Stochastic Interpolantscited by 964
  23. 2023Training Diffusion Models with Reinforcement Learningcited by 939
  24. 2022STaR: Bootstrapping Reasoning With Reasoningcited by 895
  25. 2024Simple and Effective Masked Diffusion Language Modelscited by 800
  26. 2023Diffusion Model Alignment Using Direct Preference Optimizationcited by 767
  27. 2024SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformerscited by 707
  28. 2024RDT-1B: a Diffusion Foundation Model for Bimanual Manipulationcited by 670
  29. 2022Compositional Visual Generation with Composable Diffusion Modelscited by 619
  30. 2022Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learningcited by 543
  31. 2022GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generationcited by 511
  32. 2022Is Conditional Generative Modeling all you need for Decision-Making?cited by 504
  33. 2023Learning Interactive Real-World Simulatorscited by 440
  34. 2023Universal Guidance for Diffusion Modelscited by 426
  35. 2024OminiControl: Minimal and Universal Control for Diffusion Transformercited by 334
  36. 2023Beyond Human Data: Scaling Self-Training for Problem-Solving with Language Modelscited by 325
  37. 2021Score-based diffusion models for accelerated MRIcited by 308
  38. 2023Common Diffusion Noise Schedules and Sample Steps are Flawedcited by 308

50 most cited of 80 All 80 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 17; 2026: 31.

48 papers.

Sort:

By area:

  1. 2025Reasoning with Sampling: Your Base Model is Smarter Than You Thinkcited by 84
  2. 2026Inference-time Physics Alignment of Video Generative Models with Latent World Modelscited by 32
  3. 2026Meta Flow Maps enable scalable reward alignmentcited by 32
  4. 2026Diamond Maps: Efficient Reward Alignment via Stochastic Flow Mapscited by 26
  5. 2025GLASS Flows: Transition Sampling for Alignment of Flow and Diffusion Modelscited by 17
  6. 2026Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignmentcited by 15
  7. 2025TR2-D2: Tree Search Guided Trajectory-Aware Fine-Tuning for Discrete Diffusioncited by 13
  8. 2026World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetrycited by 13
  9. 2025Learn to Guide Your Diffusion Modelcited by 12
  10. 2025RAPO++: Cross-Stage Prompt Optimization for Text-to-Video Generation via Data Alignment and Test-Time Scalingcited by 9
  11. 2025Test-Time Graph Search for Goal-Conditioned Reinforcement Learningcited by 8
  12. 2025Inference-Time Scaling of Diffusion Language Models via Trajectory Refinementcited by 7
36 more
  1. 2025Effective Test-Time Scaling of Discrete Diffusion through Iterative Refinementcited by 7
  2. 2026From Scale to Speed: Adaptive Test-Time Scaling for Image Editingcited by 7
  3. 2026Training-Free Adaptation of Diffusion Models via Doob's h-Transformcited by 6
  4. 2026Proprio: Latent Self-Scoring and Inference-Time Refinement for Physically Plausible Video Generationcited by 6
  5. 2026World Action Planner: Generalizable Decision-Making with Action-Conditioned World Modelscited by 6
  6. 2026Iterative Refinement Improves Compositional Image Generationcited by 4
  7. 2026VIGOR: VIdeo Geometry-Oriented Reward for Temporal Generative Alignmentcited by 4
  8. 2025Inference-time Scaling for Diffusion-based Audio Super-resolutioncited by 3
  9. 2026ILRR: Inference-Time Steering Method for Masked Diffusion Language Modelscited by 3
  10. 2026Aligning Few-Step Generative Models by Amortizing Sample-based Variational Inferencecited by 3
  11. 2026Temporal Backtracking Search for Test-time Generative Video Reasoningcited by 3
  12. 2025Navigating the Exploration-Exploitation Tradeoff in Inference-Time Scaling of Diffusion Modelscited by 2
  13. 2025IS-Diff: Improving Diffusion-Based Inpainting with Better Initial Seedcited by 2
  14. 2025Tree Reward-Aligned Search for TReASURe in Masked Diffusion Language Modelscited by 2
  15. 2025Diffusion Alignment as Variational Expectation-Maximizationcited by 2
  16. 2026Entropy Aware Reward Guidance for Diffusion Language Model Alignmentcited by 2
  17. 2026Compression as Adaptation: Implicit Visual Representation with Diffusion Foundation Modelscited by 2
  18. 2026S3: Stratified Scaling Search for Test-Time in Diffusion Language Modelscited by 2
  19. 2026EPIC: Efficient Predicate-Guided Inference-Time Control for Compositional Text-to-Image Generationcited by 2
  20. 2026Inference-Time Scaling in Diffusion Models through Iterative Partial Refinementcited by 2
  21. 2026Stitched Value Model for Diffusion Alignmentcited by 2
  22. 2025Ctrl-Z Sampling: Scaling Diffusion Sampling with Controlled Random Zigzag Explorationscited by 1
  23. 2025Rethinking Test Time Scaling for Flow-Matching Generative Modelscited by 1
  24. 2025Highly Efficient Test-Time Scaling for T2I Diffusion Models with Text Embedding Perturbationcited by 1
  25. 2026(1D) Ordered Tokens Enable Efficient Test-Time Searchcited by 1
  26. 2026Inference-Time Scaling of Diffusion Models via Progressive Seed Pruningcited by 1
  27. 2026FluxLite: Inference-Time Proposal Control for Discrete Diffusion Modelscited by 1
  28. 2025Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-Resolutioncited by 0
  29. 2026Plan First, Diffuse Later: Extrinsic Graph Guidance for Long-Horizon Diffusion Planningcited by 0
  30. 2026Explicit Critic Guidance for Aligning Diffusion Modelscited by 0
  31. 2026The Singularity Space: A Generative Diffusion Framework for Signal Representationcited by 0
  32. 2026MODUS: Decoder-Only Any-to-Any Modeling of Diverse Modalitiescited by 0
  33. 2026SwiftExplorer: Training-free Diffusion Model Alignment with Swift Diversity Explorationcited by 0
  34. 2026Leveraging Inference-Time Compute for Diffusion Models via Global Scheduling of Denoising Trajectoriescited by 0
  35. 2026Bridge of Psi's: Quantum Circuit Optimization with Schr"odinger Bridgescited by 0
  36. 2026AutoRef: Harness Optimization for Agentic Multi-Reference Image Generationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Dynamic Search for Inference-Time Alignment in Diffusion Models
  2. 2025Inference-Time Scaling of Discrete Diffusion Models via Importance Weighting and Optimal Proposal Design
  3. 2024An Overview of Diffusion Models: Applications, Guided Generation, Statistical Rates and Optimization
  4. 2025Diffusion Tree Sampling: Scalable inference-time alignment of diffusion models
  5. 2025Diffusion Language Model Inference with Monte Carlo Tree Search
  6. 2026Simple Approximation and Derivative Free Inference-Time Scaling for Diffusion Models via Sequential Monte Carlo on Path Measures
  7. 2025Inference-Time Alignment in Diffusion Models with Reward-Guided Generation: Tutorial and Review
  8. 2025RNE: plug-and-play diffusion inference-time control and energy-based training