AI RESEARCH
AutoRubric: Rubric-Based Generative Rewards for Faithful Multimodal Reasoning
arXiv CS.CL
•
ArXi:2510.14738v2 Announce Type: replace Multimodal large language models (MLLMs) have rapidly advanced from perception tasks to complex multi-step reasoning, yet reinforcement learning with verifiable rewards (RLVR) often leads to spurious reasoning since only the final-answer correctness is rewarded. To address this limitation, we propose AutoRubric, a framework that integrates RLVR with process-level supervision through automatically collected rubric-based generative rewards.