← 研踪 PaperTrailCVF · CVPR 2026:cvf:3daed3612bb8779bfd8bfd08d5b55f9e · 记录 #1
Multimodal RewardBench 2: Evaluating Omni Reward Models for Interleaved Text and Image
来源日期（仅到年份） 2026
来源记录日期 2026
雷达观测 2026-11-05 02:24
摘要
Reward models (RMs) are essential for training large language models (LLMs), but remain underexplored for omni models that handle interleaved image and text sequences. We introduce Multimodal RewardBench 2 (MMRB2), the first comprehensive benchmark for reward models on multimodal understanding and (interleaved) generation. MMRB2 spans four tasks: text-to-image, image editing, interleaved generation, and multimodal reasoning ("thinking-with-images"), providing 1,000 expert-annotated preference pairs per task from 23 models and agents across 21 source tasks. MMRB2 is designed with: (1) practical but challenging prompts; (2) responses from state-of-the-art models, and agents; and (3) preference pairs with strong human-expert consensus, curated via an ensemble filtering strategy. Using MMRB2, we study existing judges for each subtask, including multimodal LLM-as-a-judge and models trained with human preferences. Top judges like GPT-5 and Gemini-2.5-Pro reach 66-75% accuracy, compared to >90% for humans, and outperform the commonly used GPT-4o (59% accuracy). The best performing open-source model Qwen3-VL-32B achieves similar accuracies as Gemini-2.5-Flash (64%). We also show that MMRB2 performance strongly correlates with downstream task success, and conduct an in-depth analysis that shows key areas to improve the reward models going forward.
论文发现出处
这些页面提供论文线索，不代表同行评审或质量认证。
观测到的版本
平台记录号代表本站观测到的元数据变化，不是作者修订号。来源日期仅到月份或年份时，按该月或该年首日存档。
记录 #1 · 2026 · 最新观测版本
Multimodal RewardBench 2: Evaluating Omni Reward Models for Interleaved Text and Image
Reward models (RMs) are essential for training large language models (LLMs), but remain underexplored for omni models that handle interleaved image and text sequences. We introduce Multimodal RewardBench 2 (MMRB2), the first comprehensive benchmark for reward models on multimodal understanding and (interleaved) generation. MMRB2 spans four tasks: text-to-image, image editing, interleaved generation, and multimodal reasoning ("thinking-with-images"), providing 1,000 expert-annotated preference pairs per task from 23 models and agents across 21 source tasks. MMRB2 is designed with: (1) practical but challenging prompts; (2) responses from state-of-the-art models, and agents; and (3) preference pairs with strong human-expert consensus, curated via an ensemble filtering strategy. Using MMRB2, we study existing judges for each subtask, including multimodal LLM-as-a-judge and models trained with human preferences. Top judges like GPT-5 and Gemini-2.5-Pro reach 66-75% accuracy, compared to >90% for humans, and outperform the commonly used GPT-4o (59% accuracy). The best performing open-source model Qwen3-VL-32B achieves similar accuracies as Gemini-2.5-Flash (64%). We also show that MMRB2 performance strongly correlates with downstream task success, and conduct an in-depth analysis that shows key areas to improve the reward models going forward.
查看此版本解读 →查看此版本原文 ↗