AI Paper Atlas

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

SAIL: Self-Amplified Iterative Learning for Diffusion Model Alignment with Minimal Human Feedback

Xiaoxuan He, Siming Fu, Wanli Li, Zhiyuan Li, Dacheng Yin, Kang Rong, Fengyun Rao, Bo Zhang

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 1 citations (OpenAlex)

Read the paper

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
mix generated preference data with human seed data (mixup) (ablation)SAIL · Pick-a-Pic v2 (SD1.5, Iter2)ImageReward: 0.1564 -> 0.1729Improved
mix generated preference data with human seed data (mixup) (ablation)SAIL · Pick-a-Pic v2 (SD1.5, Iter2)HPSv2: 26.55 -> 26.65Improved
select best-worst pair instead of random pair (ablation)SAIL · Pick-a-Pic v2 (SD1.5, Iter1)ImageReward: 0.1055 -> 0.1137Improved
self-reward ranking instead of Aesthetics reward model ranking (Online DPO) (ablation)SAIL · Pick-a-Pic v2 (SD1.5, Iter1)ImageReward: 0.0936 -> 0.1137Improved
self-reward ranking instead of Aesthetics reward model ranking (Online DPO) (ablation)SAIL · Pick-a-Pic v2 (SD1.5, Iter1)Aesthetics: 5.53 -> 5.46Hurt
add more self-improvement iterations (ablation)SAIL · Pick-a-Pic v2 (SD1.5)ImageReward: 0.0715 -> 0.2329Improved
add a further iteration (Iter4 vs Iter3) (ablation)SAIL · PartiPrompts (SD1.5)ImageReward: 0.3198 -> 0.3072No clear effect
use 6% of human preference data with iterative self-rewarding instead of full DiffusionDPO dataSDXL · Pick-a-Pic v2 validationHPSv2: 27.27 -> 27.32Improved
use 6% of human preference data instead of full DiffusionSPO dataSD1.5 · Pick-a-Pic v2 validationPickScore: 21.20 -> 21.00No clear effect
initialize SAIL with the full Pick-a-Pic v2 data (SAIL*)SD1.5 · Pick-a-Pic v2 validationImageReward: 0.1577 -> 0.4303 vs DiffusionSPOImproved

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