StructVRM: Aligning Multimodal Reasoning with Structured and Verifiable Reward Models

Open in new window