AI RESEARCH
Bridging Perception and Reasoning: Token Reweighting for RLVR in Multimodal LLMs
arXiv CS.CV
•
ArXi:2603.25077v1 Announce Type: new Extending Reinforcement Learning with Verifiable Rewards (RLVR) to multimodal large language models (MLLMs) faces a fundamental challenge: their responses inherently interleave perception-related tokens, which ground visual content, with reasoning-related tokens, which construct reasoning chains. These token types instantiate distinct yet interdependent capacities -- visual grounding and symbolic reasoning -- making isolated optimization insufficient.