AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Bridging Visual Representation and Reinforcement Learning from Verifiable Rewards in Large Vision-Language Models

yuhang han, Yuyang Wu, Zhengbo Jiao, Yiyu Wang, Xuyang Liu, Shaobo Wang, Hanlin xu, Xuming Hu, Linfeng Zhang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Reinforcement learning for visual reasoning · 3 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 KAWHI to GRPOGRPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen2.5-VL-7Baverage accuracy: 50.18 -> 51.00Improved
add KAWHI to DAPODAPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen2.5-VL-7Baverage accuracy: 50.79 -> 51.85Improved
add KAWHI to GSPOGSPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen2.5-VL-7Baverage accuracy: 51.40 -> 52.94Improved
add KAWHI to GRPOGRPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen3-VL-4Baverage accuracy: 51.17 -> 51.98Improved
add KAWHI to DAPODAPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen3-VL-4Baverage accuracy: 51.80 -> 53.06Improved
add KAWHI to GSPOGSPO · Math reasoning avg (MathVista, MathVerse, MathVision, WeMath), Qwen3-VL-4Baverage accuracy: 54.14 -> 55.54Improved
add KAWHI to GRPOGRPO · MathVista, Qwen2.5-VL-7Baccuracy: 68.37 -> 69.30Mixed
add KAWHI to GRPO (chart training)GRPO · CharXiv Descaccuracy: 70.1 -> 72.2Improved
add KAWHI to GRPO (chart training)GRPO · ChartMimicaccuracy: 42.8 -> 45.1Improved
use SGUF-selected regions instead of all vision tokens (ablation)GRPO+KAWHI · MathVerseaccuracy: 46.58 -> 47.56Improved
use SGUF-selected regions instead of inverse selection (ablation)GRPO+KAWHI · MathVerseaccuracy: 45.74 -> 47.56Improved
use key-query instead of key-key similarity (ablation)GRPO+KAWHI · MathVerseaccuracy: 47.08 -> 47.56Improved

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