AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Cones: Concept Neurons in Diffusion Models for Customized Generation

Zhiheng Liu, Ruili Feng, Kai Zhu, Yifei Zhang, Kecheng Zheng, Yu Liu, Deli Zhao, Jingren Zhou, Yang Cao

ICML 2023 · Image generation › Fast diffusion sampling · approach: Diffusion models as perception tools · 19 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
Use gradient-identified, shut concept neurons instead of Custom DiffusionCustomized generation · Single SubjectText-alignment: 0.352 -> 0.361Improved
Use gradient-identified, shut concept neurons instead of Custom DiffusionCustomized generation · Single SubjectImage-alignment: 0.722 -> 0.725; DreamBooth 0.731; Textual Inversion 0.744Mixed
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Two SubjectsText-alignment: 0.314 -> 0.337Improved
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Two SubjectsImage-alignment: 0.685 -> 0.698Improved
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Three SubjectsText-alignment: 0.289 -> 0.301Improved
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Three SubjectsImage-alignment: 0.669 -> 0.685Improved
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Four SubjectsText-alignment: 0.269 -> 0.285Improved
Concatenate subject-specific concept-neuron clusters rather than use Custom DiffusionMulti-subject customized generation · Four SubjectsImage-alignment: 0.632 -> 0.653Improved
Use binary shutting rather than float32, float16, or quaternary concept-neuron updates (ablation)Cones · Customized subject generationGeneration performanceNo clear effect
Fine-tune concatenated concept neurons rather than learn multi-subject neurons from scratch (ablation)Cones · Multi-subject generationGeneration performanceImproved
Concatenate concept-neuron clusters rather than use Custom Diffusion constraint optimizationTuning-free multi-subject generation · Visual quality and subject-generation accuracyGeneration qualityImproved

Later papers that compared with this method

6 later accepted papers report Cones2 as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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