cs.AIAug 9, 2026

Improving Generalization Robustness of Multimodal RLVR

Authors: Pengfei ZhouZhiwei TangXiaopeng PengChenrui ZhouLama MoukheiberYixing MaBin XuJiajun Song+6 more

Organizations: 1National University of Singapore, HPC-AI Lab · 2DAMO Academy, Alibaba Group · 3Hupan Lab · 4Zhejiang University · 6Rochester Institute of Technology · 7Georgia Institute of Technology · University of California, Berkeley · 8InfRec, Cardinal AI Lab · 9Renmin University of China · 10Hong Kong University of Science and Technology

Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) makes Multimodal Large Language Models more accurate, but the gains are brittle: simply paraphrasing a question or changing the prompt template can degrade them, which challenges reliable deployment in high-stakes scenarios like medical VQA. We trace this to two issues of the standard RL objective. First, the binary verifier conflates format with content, so the reward signal cannot tell a wrong answer apart from a misformatted one. Second, the training distribution covers only a thin slice of the real-world prompts that the model might meet at deployment, so policies that perform well on the training distribution can behave differently under unseen prompts during test. Both failures call for a robust post-training method that helps the policy cover a broader distribution of semantically equivalent prompts, and we identify two measures that help achieve this objective: separating format from semantics in the reward, and applying policy invariance across perturbed prompts with equivalent semantics. We therefore propose Prompt-Invariant RLVR (PIRL), consisting of a dynamic trinary reward and a consistency regularizer based on an embedding-space adversary. Under stress testing, PIRL's average accuracy on benchmarks drops by only 1%\le 1\%, where GRPO drops ~3%. On dynamic evaluation, PIRL also achieves the smallest performance drop.

Explore similar work

CardsList