mlsanityData to 14 Oct 2026

Feedback Guidance of Diffusion Models

arXiv 2506.06085 ↗First posted 6 Jun 2025cs.CVcited by 16

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

While Classifier-Free Guidance (CFG) has become standard for improving sample fidelity in conditional diffusion models, it can harm diversity and induce memorization by applying constant guidance regardless of whether a particular sample needs correction. We propose FeedBack Guidance (FBG), which uses a state-dependent coefficient to self-regulate guidance amounts based on need. Our approach is derived from first principles by assuming the learned conditional distribution is linearly corrupted by the unconditional distribution, contrasting with CFG's implicit multiplicative assumption. Our scheme relies on feedback of its own predictions about the conditional signal informativeness to adapt guidance dynamically during inference, challenging the view of guidance as a fixed hyperparameter. The approach is benchmarked on ImageNet512x512, where it significantly outperforms Classifier-Free Guidance and is competitive to Limited Interval Guidance (LIG) while benefitting from a strong mathematical framework. On Text-To-Image generation, we demonstrate that, as anticipated, our approach automatically applies higher guidance scales for complex prompts than for simpler ones and that it can be easily combined with existing guidance schemes such as CFG or LIG.

Key facts. Cited by 16 papers in the map (data through 14 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 43 of its 48 references that are in the map. First posted on arXiv 6 Jun 2025.

Builds on

43 of 48 references.

Closest ancestors · 33

Same direction, newest first

  1. 2025DDPM Score Matching and Distribution Learning7 Apr
  2. 2025Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps16 Jan
  3. 2024The Superposition of Diffusion Models Using the Ito Density Estimator23 Dec
  4. 2024Simple Guidance Mechanisms for Discrete Diffusion Models13 Dec
  5. 2024Self-Guidance: Boosting Flow and Diffusion Generation on Their Own8 Dec
  6. 2024Diffusion Models as Cartoonists: The Curious Case of High Density Regions2 Nov
  7. 2024Rectified Diffusion Guidance for Conditional Generation24 Oct
  8. 2024Dynamic Negative Guidance of Diffusion Models18 Oct
  9. 2024Conditional Image Synthesis with Diffusion Models: A Survey28 Sep
  10. 2024TFG: Unified Training-Free Guidance for Diffusion Models24 Sep
  11. 2024What does guidance do? A fine-grained analysis in a simple setting19 Sep
  12. 2024Classifier-Free Guidance is a Predictor-Corrector16 Aug
21 more
  1. 2024CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models12 Jun
  2. 2024Guiding a Diffusion Model with a Bad Version of Itself4 Jun
  3. 2024Analysis of Classifier-Free Guidance Weight Schedulers19 Apr
  4. 2024Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models11 Apr
  5. 2024Rethinking the Spatial Inconsistency in Classifier-Free Diffusion Guidance8 Apr
  6. 2024Understanding and Improving Training-free Loss-based Diffusion Guidance19 Mar
  7. 2024Critical windows: non-asymptotic theory for feature emergence in diffusion models3 Mar
  8. 2024Dynamical Regimes of Diffusion Models28 Feb
  9. 2024A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data26 Feb
  10. 2023Analyzing and Improving the Training Dynamics of Diffusion Models5 Dec
  11. 2023The statistical thermodynamics of generative diffusion models: Phase transitions, symmetry breaking and critical instability26 Oct
  12. 2023CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling26 Oct
  13. 2023Spontaneous Symmetry Breaking in Generative Diffusion Models31 May
  14. 2022Scalable Diffusion Models with Transformers19 Dec
  15. 2022Classifier-Free Diffusion Guidance26 Jul
  16. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  17. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  18. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  19. 2021Maximum Likelihood Training of Score-Based Diffusion Models22 Jan
  20. 2020Denoising Diffusion Implicit Models6 Oct
  21. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 10

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  4. 2022Flow Matching for Generative Modelingcited by 5,702
  5. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  6. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 944
  7. 2023Exposing flaws of generative model evaluation metrics and their unfair treatment of diffusion modelscited by 205
  8. 2023Guided Flows for Generative Modeling and Decision Makingcited by 101
  9. 2025Training-Free Guidance Beyond Differentiability: Scalable Path Steering with Tree Search in Diffusion and Flow Modelscited by 24
  10. 2025Training-Free Safe Denoisers for Safe Use of Diffusion Modelscited by 21

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 11.

16 papers.

Sort:

By area:

  1. 2025Learn to Guide Your Diffusion Modelcited by 14
  2. 2026The Entropic Signature of Class Speciation in Diffusion Modelscited by 9
  3. 2025VSF: Simple, Efficient, and Effective Negative Guidance in Few-Step Image Generation Models By Value Sign Flipcited by 4
  4. 2026Emergence of Distortions in High-Dimensional Guided Diffusion Modelscited by 4
  5. 2026Momentum Guidance: Plug-and-Play Guidance for Flow Modelscited by 3
  6. 2025MAMBO-G: Magnitude-Aware Mitigation for Boosted Guidancecited by 1
  7. 2025Highly Efficient Test-Time Scaling for T2I Diffusion Models with Text Embedding Perturbationcited by 1
  8. 2026When Does High-CFG Diffusion Inversion Fail? A Controlled Study of Prompt--Latent Interactionscited by 1
  9. 2026PathGuide: Dynamic Classifier-Free Guidance via On-Policy Transport Alignmentcited by 1
  10. 2025Physics-Guided Null-Space Diffusion with Sparse Masking for Corrective Sparse-View CT Reconstructioncited by 0
  11. 2026A Priori Sampling of Transition States with Guided Diffusioncited by 0
  12. 2026Diverse Sampling in Diffusion Models with Marginal Preserving Particle Guidancecited by 0
4 more
  1. 2026Inline Critic Steers Image Editingcited by 0
  2. 2026Spectral Guidance for Flexible and Efficient Control of Diffusion Modelscited by 0
  3. 2026Adversarial Learning of Classifier-Free Guidance Schedulescited by 0
  4. 2026DynG-Diff: A State-Aware Dynamic Guidance Diffusion Framework for Probabilistic Time Series Forecastingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2023Adaptive Guidance: Training-free Acceleration of Conditional Diffusion Models
  2. 2025Stage-wise Dynamics of Classifier-Free Guidance in Diffusion Models
  3. 2026Revisiting Classifier-Free Guidance Methods in Latent Diffusion Models
  4. 2023Bridging the Gap: Addressing Discrepancies in Diffusion Model Training for Classifier-Free Guidance
  5. 2026Guiding a Diffusion Model by Swapping Its Tokens
  6. 2025REG: Rectified Gradient Guidance for Conditional Diffusion Models
  7. 2026Opening the Black Box of Classifier-Free Guidance via Information Bottleneck
  8. 2026C2FG: Control Classifier-Free Guidance via Score Discrepancy Analysis