MMRB-2

MMRB-2 is a benchmark dataset used to evaluate multimodal judgment capabilities, specifically focusing on tasks related to image editing and multimodal reasoning.

Papers using MMRB-2 (3)

AutoRubric-T2I: Robust Rule-Based Reward Model for Text-to-Image Alignment2026MJ1: Multimodal Judgment via Grounded Verification2026Joint Reward Modeling: Internalizing Chain-of-Thought for Efficient Visual Reward Models2026