AI Paper Atlas

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

Diffusion Negative Preference Optimization Made Simple

Joshua Tian Jin Tee, Hee Suk Yoon, Sunjae Yoon, Tri Ton, Chang D. Yoo

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 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 Diff-SNPO instead of Diff-DPOSD1.5 · Pick-a-Pic v2 testPickScore: 21.01 -> 22.24Improved
use single-model Diff-SNPO instead of dual-model Diff-NPOSD1.5 · Pick-a-Pic v2 testHPSv2: 26.92 -> 27.23Improved
use Diff-SNPO instead of Diff-NPO on SDXLSDXL · Pick-a-Pic v2 testAesthetic Score: 5.9449 -> 5.8129Hurt
use Diff-SNPO instead of Diff-NPO on SDXLSDXL · Pick-a-Pic v2 testHPSv2: 28.30 -> 28.33No clear effect
add interpolation (lambda 0.9) to avoid naive single-model NPO (lambda 1.0) (ablation)Diff-SNPO · Pick-a-Pic v2 (SD1.5)HPSv2 and AestheticImproved
vary lambda around 0.9 (ablation)Diff-SNPO · Pick-a-Pic v2 (SD1.5)HPSv2 and AestheticNo clear effect
vary beta from 1000 to 3000 (ablation)Diff-SNPO · Pick-a-Pic v2 (SD1.5)HPSv2: 27.30 -> 27.23No clear effect
avoid weight merging by using single modelDiff-SNPO vs Diff-NPO · Pick-a-Pic v2 testnegative implicit accuracy (%): 52.34 -> 57.45Improved
single-model design vs dual-model trainingDiff-SNPO · SD1.5 training costrelative training speed: 1.00x -> 2.00xImproved
single-model design vs dual-model inferenceDiff-SNPO · SD1.5 inferencethroughput (img/s): 0.27 -> 0.48Improved
swap sampler from DDIM to UniPC (ablation)Diff-SNPO · Pick-a-Pic v2 (SD1.5)HPSv2: 27.23 -> 27.25No clear effect
use CHATS instead of Diff-SNPO on SD1.5CHATS · Pick-a-Pic v2 (SD1.5)Aesthetic Score: 5.6258 -> 5.6845Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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