DEEPO: Dual-Entropy Enhanced Policy Optimization for Hallucination in MLLMs
A new reinforcement learning framework designed to reduce hallucinations in multimodal models through dual entropy.
Reinforcement learning (RL) is widely used to sharpen reasoning in multimodal large language models (MLLMs), yet its effect on hallucination is uneven. We trace this to two weak points in the \emph{correction chain} from reward to parameter update. At the rollout level, hard quer...
ArXiv CS.AI
Source
Generated insight by OnAirToday AI
DEEPO addresses the critical instability in reinforcement learning for multimodal models by refining how reward signals translate into structural parameter updates. By mitigating hallucination through dual-entropy optimization, this approach creates a more reliable framework for models interpreting complex visual and textual data.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.