mlsanityData to 14 Oct 2026

Interpretable Diffusion Models with B-cos Networks

arXiv 2507.03846 ↗First posted 5 Jul 2025cs.CV cs.LGcited by 0

On the map: Diffusion models → Image generation, text-to-image and editing

Abstract (arXiv)

Text-to-image diffusion models generate images by iteratively denoising random noise, conditioned on a prompt. While these models have enabled impressive progress in image generation, they often fail to accurately reflect all semantic information described in the prompt -- failures that are difficult to detect automatically. In this work, we introduce a diffusion model architecture built with B-cos modules that offers inherent interpretability. Our approach provides insight into how individual prompt tokens affect the generated image by producing explanations that highlight the pixel regions influenced by each token. We demonstrate that B-cos diffusion models can produce high-quality images while providing meaningful insights into prompt-image alignment.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models. Builds on 21 of its 25 references that are in the map. First posted on arXiv 5 Jul 2025.

Builds on

21 of 25 references.

Closest ancestors · 4

Same direction, newest first

  1. 2024On Discrete Prompt Optimization for Diffusion Models27 Jun
  2. 2023Self-Discovering Interpretable Diffusion Latent Directions for Responsible Text-to-Image Generation28 Nov
  3. 2023Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models25 May
  4. 2022What the DAAM: Interpreting Stable Diffusion Using Cross Attention10 Oct

Other references · 17

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  3. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  5. 2020Denoising Diffusion Implicit Modelscited by 9,781
  6. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  7. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  8. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  9. 2021CLIPScore: A Reference-free Evaluation Metric for Image Captioningcited by 2,424
  10. 2019Full-Gradient Representation for Neural Network Visualizationcited by 191
  11. 2023Diffuse, Attend, and Segment: Unsupervised Zero-Shot Segmentation using Stable Diffusioncited by 131
  12. 2022B-cos Networks: Alignment is All We Need for Interpretabilitycited by 88
5 more
  1. 2023Diffusion Model is Secretly a Training-free Open Vocabulary Semantic Segmentercited by 76
  2. 2023Interpretable Diffusion via Information Decompositioncited by 37
  3. 2023B-cos Alignment for Inherently Interpretable CNNs and Vision Transformerscited by 31
  4. 2024B-cosification: Transforming Deep Neural Networks to be Inherently Interpretablecited by 12
  5. 2024B-Cos Aligned Transformers Learn Human-Interpretable Featurescited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Reusing Computation in Text-to-Image Diffusion for Efficient Generation of Image Sets
  2. 2025Emergence and Evolution of Interpretable Concepts in Diffusion Models
  3. 2024DiffusionPID: Interpreting Diffusion via Partial Information Decomposition
  4. 2023CONFORM: Contrast is All You Need For High-Fidelity Text-to-Image Diffusion Models
  5. 2025Patronus: Interpretable Diffusion Models with Prototypes
  6. 2023Reverse Stable Diffusion: What prompt was used to generate this image?
  7. 2023Semantic Guidance Tuning for Text-To-Image Diffusion Models
  8. 2024InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization