AI Paper Atlas

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

UniVerse: Empower Unified Generation with Reasoning and Knowledge

Kaiyue Sun, Weiyang Jin, Chengqi Duan, Rongyao Fang, Xian Liu, Yuwei Niu, Chunwei Wang, Aoxue Li, Xihui Liu

CVPR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters

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 CoT injection to UniVerse training (ablation)Bagel · R2I-Bench, no thinkingoverall score: 0.40 -> 0.44Improved
train on UniVerse without CoTBagel · WISE, no thinkingoverall score: 0.52 -> 0.55Improved
add CoT injection to UniVerse training (ablation)Bagel · WISE, no thinkingoverall score: 0.55 -> 0.56No clear effect
train on UniVerse with CoT injectionBagel · R2I-Bench, no thinkingoverall score: 0.36 -> 0.44Improved
add CoT injection to UniVerse training (ablation)Bagel · R2I-Bench Causal, no thinkingscore: 0.32 -> 0.69Improved
add CoT injection to UniVerse training (ablation)Bagel · R2I-Bench Common Sense (CS), no thinkingscore: 0.42 -> 0.38Hurt
add CoT injection to UniVerse training (ablation)Bagel · WISE, thinking enabledoverall score: 0.74 -> 0.77Improved
add CoT injection to UniVerse training (ablation)Bagel · R2I-Bench, thinking enabledoverall score: 0.51 -> 0.54Improved
enable thinking mode at evaluationBagel · WISEoverall score: 0.52 -> 0.70Improved
train on all UniVerse data instead of a single category (ablation)Bagel w/ CoT · WISE, thinking enabledoverall score: 0.74 -> 0.77Improved
train on Reasoning subset (largest category) (ablation)Bagel w/ CoT · WISE, thinking enabledoverall score: 0.70 -> 0.74Improved
keep salvaged samples (realigned after filtering) in arithmetic data (ablation)Bagel w/ CoT · WISE, thinking enabledoverall score: 0.71 -> 0.72Improved

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