AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

UnHype: CLIP-Guided Hypernetworks for Dynamic LoRA Unlearning

Piotr Wójcik, Maksym Petrenko, Wojciech Gromski, Przemysław Spurek, Maciej Zieba

ICML 2026 · Image generation › Text-to-image diffusion · approach: Concept erasure in diffusion models · 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
Use CLIP-conditioned hypernetwork (UnHype) instead of per-concept ESD-u for object erasureSD v1.4 object erasure · CIFAR-10 object erasure (avg of 3 classes)H_o: 86.96 -> 93.85Improved
Apply UnHype to nudity erasure instead of leaving SD v1.4 unmodifiedSD v1.4 · I2P NudeNet totaltotal detections: 743 -> 8Improved
Apply UnHype nudity erasure on Flux instead of unmodified modelFlux.1 [dev] · I2P NudeNet totaltotal detections: 605 -> 32Improved
Use UnHype instead of EraseAnything for Flux nudity erasureFlux.1 [dev] · I2P NudeNet totaltotal detections: 199 -> 32Improved
Use UnHype for single-model 100-celebrity erasure instead of TRCESD v1.4 · 100-celebrity erasureH_o: 89.85 -> 92.48Improved
Use UnHype instead of AdvUnlearn against UnlearnDiffAtkSD v1.4 nudity · UnlearnDiffAtk ASRattack success rate (%): 3.40 -> 0.00Improved
Use UnHype instead of AdvUnlearn against Ring-A-BellSD v1.4 nudity · Ring-A-Bell ASRattack success rate (%): 0.00 -> 1.05Hurt
Use UnHype instead of ESD-u for 10-class Imagenette removalSD v1.4 · ImagenetteCLIP score: 22.52 -> 30.61Improved
Use UnHype instead of ScaPre on ImageNet-Confuse5SD v1.4 · ImageNet-Confuse5Overall Acc: 84.3 -> 87.0Improved
Use aggressive erasers UCE/RECE instead of UnHype on Confuse5SD v1.4 · ImageNet-Confuse5Preserve Acc: 83.3 -> 5.6Hurt
Increase LoRA rank from 1 to 9 (ablation)UnHype · Imagenettetotal ResNet-50 accuracy: 0.06 -> 0.04No clear effect
Increase LoRA rank from 1 to 12 (ablation)UnHype · ImagenetteCLIP score: 30.61 -> 30.61No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

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