AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Diffusion Models Preferentially Memorize Prototypical Examples or: Why Does My Diffusion Model Love Slop?

Marta Aparicio Rodriguez, Anastasia Borovykh, Grigorios A. Pavliotis, Daniel James Korchinski

ICML 2026 · Image generation › Fast diffusion sampling · approach: Diffusion models as perception tools · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
raise sample log-likelihood (more common samples)diffusion model on RHM data · RHMtime to memorization: negative linear relation between log-likelihood and time to memorizationImproved
increase Zipf exponent alpha (thinner tails)diffusion model on RHM data · RHM, width 4096mean time to memorization: mean memorization time decreases with increasing alphaImproved
move Zipf variation to more abstract level (smaller l_z)diffusion model on RHM data · RHM, width 1024, alpha=2time to memorization: l_z=0 delays memorization versus l_z=3Hurt
increase model widthdiffusion model on RHM data · RHMmemorization speed: 4096 model >95% memorization in 4000 epochs vs 256 model barely 10% after 100000 epochsImproved
add Zipf variation at root level l_z=0diffusion model on RHM data · RHMpreferential memorization of common samplesNo clear effect
decrease model sizediffusion model on RHM data · RHMpartial memorization of frequent rule: more pronounced in the width 256 modelHurt
decrease model size (109M to 63M parameters)DDPM on CelebA · CelebA 10,000 imagesslop phase length: longer slop phase for 63M modelHurt
early stopping at train/test loss divergencediffusion model on RHM data · RHM, l_z=3overproduction of common subfeatures: test loss increase generally precedes KL increaseImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy