AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Local Mechanisms of Compositional Generalization in Conditional Diffusion

Arwen Bradley

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
enforce local conditional scores with a patch-local architecture (Exp.2L vs Exp.2) (ablation)single-object grid-conditioned CLEVR model, trained on 1-3 objects · Location-conditioned CLEVRK_max: 2 -> 9Improved
enforce local architecture (Exp.2L vs Exp.2), trained on 1 object (ablation)single-object grid-conditioned model · Location-conditioned CLEVR, trained on 1 objectK_max: 1 -> 5Improved
enforce local architecture (Exp.2L vs Exp.2), trained on 1-5 objects (ablation)single-object grid-conditioned model · Location-conditioned CLEVR, trained on 1-5 objectsK_max: 3 -> 10Improved
enforce local architecture (Exp.3L vs Exp.3), trained on 1-3 objects (ablation)list-conditioned model · Location-conditioned CLEVR, trained on 1-3 objectsK_max: 3 -> 9Improved
label all object locations with grid conditioner instead of a single object (Exp.1 vs Exp.2), 1-3 objectsgrid-conditioned CLEVR model · Location-conditioned CLEVR, trained on 1-3 objectsK_max: 2 -> 7Improved
use grid-style instead of list-style conditioner (Exp.1 vs Exp.3), 1-3 objectslocation-conditioned CLEVR model · Location-conditioned CLEVR, trained on 1-3 objectsK_max: 3 -> 7Improved
train on more objects (1 to 1-3 to 1-5) for Exp.1grid-conditioned model · Location-conditioned CLEVRK_max: 1 -> 7 -> 10Improved
train on more objects (1-3 to 1-5) for the list-conditioned modelExp.3 · Location-conditioned CLEVRK_max: 3 -> 6Improved
feature-space instead of pixel-space for disentanglement of conceptsSDXL · SDXL hidden activations, 24 promptsdisentanglement ratio: 1.42 -> 1.4--1.6Improved

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