AI Paper Atlas

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

Multimodal RewardBench 2: Evaluating Omni Reward Models for Interleaved Text and Image

Yushi Hu, Reyhane Askari-Hemmat, Melissa Hall, Emily Dinan, Luke Zettlemoyer, Marjan Ghazvininejad

CVPR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 21 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
preference-train Qwen2.5-VL-7B as EditRewardQwen2.5-VL-7B · MMRB2 image editing (single-image subset)accuracy: 57.8 -> 67.2Improved
preference-train Qwen2.5-VL-7B as UnifiedRewardQwen2.5-VL-7B · MMRB2 text-to-imageaccuracy: 50.4 -> 59.8Improved
preference-train Qwen2.5-VL-7B as UnifiedRewardQwen2.5-VL-7B · MMRB2 multimodal reasoning (text-only-output subset)accuracy: 53.7 -> 55.1Improved
add preference training to CLIPScore (ImageReward)CLIPScore · MMRB2 text-to-imageaccuracy: 51.0 -> 54.0Improved
add preference training to CLIPScore (HPSv2)CLIPScore · MMRB2 text-to-imageaccuracy: 51.0 -> 54.7Improved
add preference training to CLIPScore (PickScore)CLIPScore · MMRB2 text-to-imageaccuracy: 51.0 -> 58.6Improved
use larger open MLLM judge Qwen3-VL-32B instead of preference-trained HPSv3 on text-to-imageQwen3-VL-32B · MMRB2 text-to-imageaccuracy: 60.2 -> 64.1Improved
scale Gemma 3 judge from 4B to 27BGemma 3 · MMRB2 (average of 4 tasks)accuracy: 50.7 -> 57.3Improved
scale Qwen2.5-VL judge from 7B to 72BQwen2.5-VL · MMRB2 (average of 4 tasks)accuracy: 50.9 -> 59.0Improved
evaluate different-model pairs instead of same-model pairs (ablation)Gemini 3 Pro judge · MMRB2 image generationaccuracy: 70.4 -> 79.7Improved
majority vote over K=9 samples instead of K=1 (ablation)Gemini 2.5 Pro judge · MMRB2 (average of 4 tasks)accuracy: 71.3 -> 72.5Improved
majority vote over more samples K (ablation)Qwen3-VL judges · MMRB2 (average of 4 tasks)accuracyNo 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