AI Paper Atlas

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

Leveraging Verifier-Based Reinforcement Learning in Image Editing

Hanzhong Guo, Jie Wu, Jie Liu, Yu Gao, Zilyu Ye, Linxiao Yuan, Xionghui Wang, Yizhou Yu, Weilin Huang

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 16 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 GCPO reinforcement stage after SFT (ablation)Edit-RRM 7B · Internal reward benchmarkaccuracy: 75.4% -> 82.2%Improved
add GCPO reinforcement stage after SFT (ablation)Edit-RRM 3B · Internal reward benchmarkaccuracy: 69.3% -> 72.0%Improved
add GCPO reinforcement stage after SFT (ablation)Edit-RRM 7B · EditRewardBenchaccuracy: 73.3% -> 78.2%Improved
add external Verify filtering to Think-only SFT data (ablation)Edit-RRM 7B · Internal reward benchmarkaccuracy: 68.9% -> 75.4%Improved
use principle-decomposition SFT data instead of VIESCORE prompts (ablation)Edit-RRM 7B · Internal reward benchmarkaccuracy: 68.3% -> 75.4%Improved
scale reward model from 3B to 7B (ablation)Edit-RRM · Internal reward benchmarkaccuracy (Think+Verify+GCPO): 72.0% -> 82.2%Improved
use verifier-based RRM instead of holistic scorer EditScore-7BEdit-RRM SFT · EditRewardBenchaccuracy: 65.9% -> 73.3%Improved
use RL-RRM (7B) as GRPO reward for FLUX.KontextFLUX.Kontext · GEdit-Bench-ENoverall O: 5.77 -> 6.24Improved
use RL-RRM (7B) as GRPO reward for FLUX.KontextFLUX.Kontext · GEdit-Bench-ENSC: 6.27 -> 6.86Improved
use RL-RRM (7B) as GRPO reward for FLUX.KontextFLUX.Kontext · GEdit-Bench-ENPQ: 7.25 -> 7.20No clear effect
use GCPO-refined RRM instead of SFT-only RRM as reward (7B) (ablation)FLUX.Kontext · GEdit-Bench-ENoverall O: 6.20 -> 6.24Improved
use GCPO-refined RRM instead of SFT-only RRM as reward (3B) (ablation)FLUX.Kontext · GEdit-Bench-ENoverall O: 5.63 -> 6.10Improved

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