# Origins of Creativity in Attention-Based Diffusion Models

arXiv: https://arxiv.org/abs/2506.17324 · HTML: https://mlsanity.com/paper/2506.17324/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 6 of its 9 references that are in the map.
- First posted on arXiv 18 Jun 2025.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Builds on (most cited first, 6 of 6)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Deep Unsupervised Learning using Nonequilibrium Thermodynamics](https://mlsanity.com/paper/1503.03585/) (2015)
- [Categorical Reparameterization with Gumbel-Softmax](https://mlsanity.com/paper/1611.01144/) (2016)
- [Dynamical Regimes of Diffusion Models](https://mlsanity.com/paper/2402.18491/) (2024)
- [An analytic theory of creativity in convolutional diffusion models](https://mlsanity.com/paper/2412.20292/) (2024)
- [A Good Score Does not Lead to A Good Generative Model](https://mlsanity.com/paper/2401.04856/) (2024)

## Built on by (most cited first, 4 of 4)

- [A Random Matrix Theory Perspective on the Consistency of Diffusion Models](https://mlsanity.com/paper/2602.02908/) (2026)
- [An exact information theory of generalization phase transitions in Bayesian diffusion models](https://mlsanity.com/paper/2607.08041/) (2026)
- [The two clocks and the innovation window: When and how generative models learn rules](https://mlsanity.com/paper/2605.10019/) (2026)
- [Representation Gap: Explaining the Unreasonable Effectiveness of Neural Networks from a Geometric Perspective](https://mlsanity.com/paper/2605.21692/) (2026)

## Abstract (arXiv)

As diffusion models have become the tool of choice for image generation and as the quality of the images continues to improve, the question of how `creativity' originates in diffusion has become increasingly important. The score matching perspective on diffusion has proven particularly fruitful for understanding how and why diffusion models generate images that remain plausible while differing significantly from their training images. In particular, as explained in (Kamb & Ganguli, 2024) and others, e.g., (Ambrogioni, 2023), theory suggests that if our score matching were optimal, we would only be able to recover training samples through our diffusion process. However, as shown by Kamb & Ganguli, (2024), in diffusion models where the score is parametrized by a simple CNN, the inductive biases of the CNN itself (translation equivariance and locality) allow the model to generate samples that globally do not match any training samples, but are rather patch-wise `mosaics'. Notably, however, this theory does not extend to describe the role of self-attention in this process. In this work, we take a preliminary step in this direction to extend this theory to the case of diffusion models whose score is parametrized by a CNN with a final self-attention layer. We show that our theory suggests that self-attention will induce a globally image-consistent arrangement of local features beyond the patch-level in generated samples, and we verify this behavior empirically on a carefully crafted dataset.