AI Paper Atlas

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

Stable Preference: Redefining training paradigm of human preference model for Text-to-Image Synthesis

Hanting Li, Hongjing Niu, Feng Zhao

ECCV 2024 · 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
add anti-interference loss (ablation)CLIP-H on ImageReward · ImageReward testpair selection accuracy: 63.9 -> 64.5Improved
use stable preference two-step paradigm instead of cross-entropy (ablation)CLIP-H on ImageReward · ImageReward testpair selection accuracy: 63.9 -> 66.2Improved
combine anti-interference loss with stable preference (ablation)CLIP-H · ImageReward testpair selection accuracy: 66.2 -> 66.8Improved
use stable preference on CLIP-L (ablation)CLIP-L · ImageReward testpair selection accuracy: 63.1 -> 66.3Improved
weight step-2 loss toward larger score-gap pairs instead of uniform (ablation)Stable Preference · HPD v2pair selection accuracy: 80.1 -> 80.7Improved
weight step-2 loss toward similar pairs instead of large-gap pairs (ablation)Stable Preference · HPD v2pair selection accuracy: 80.7 -> 79.5Hurt
train CLIP-H with stable preference vs cross-entropy, evaluated cross-domainCLIP-H · HPD v2 testpair selection accuracy: 74.8 -> 80.7Improved
initialize CLIP-H from HPS v2 checkpointStable Preference · ImageReward testpair selection accuracy: 66.8 -> 68.0Improved
apply horizontal flip and center crop at test timeSP (CLIP-H) vs HPS v2 · HPD v2pair selection accuracy: HPS v2 83.3 -> 81.7 vs SP 80.7 -> 80.7Improved

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