AI Paper Atlas

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

Diffusion Blend: Inference-Time Multi-Preference Alignment for Diffusion Models

Min Cheng, Fatemeh Doudi, Dileep Kalathil, Mohammad Ghavamzadeh, Panganamala Kumar

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion

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
blend backward diffusion of reward-specific fine-tuned models (DB-MPA) instead of averaging weights (Rewarded Soup)Stable Diffusion v1.5 · Short-DrawBench w=0.5weighted reward: outperforms RS by 3.92xImproved
sample LoRA adapters per denoising step (DB-MPA-LS) instead of evaluating all models (ablation)DB-MPA · Short-DrawBench w=0.5weighted reward / time: 0.42 -> 0.39; 11.11 -> 5.64 sec/imgNo clear effect
increase number of rewards from 2 to 4 (ablation)DB-MPA-LS vs DB-MPA · Multi-rewardDB/LS gain ratio: 1.02, 1.15, 1.08 for 2, 3, 4 rewardsNo clear effect
increase number of rewards from 2 to 4 (ablation)Rewarded Soup · Multi-rewardDB/RS gain ratio: 3.3, 6.4, 8.2Hurt
blend pretrained and fine-tuned drift with factor lambda (DB-KLA)Fine-tuned SD v1.5 (alpha=0.1) · Short-DrawBenchaverage rewardImproved
use lambda > 1 in DB-KLADB-KLA · Short-DrawBenchsemantic accuracyImproved
add conflicting JPEG compressibility rewardDB-MPA · Short-DrawBenchreward ParetoNo clear effect
scale to GenEval and SDXLDB-MPA · GenEval 700 held-out promptsPareto frontImproved
use DB-MPA-LS sampling for KL control (ablation)DB-KLA · KL alignmentfeasibilityNo clear effect

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