MMRAG-RFT: Two-stage Reinforcement Fine-tuning for Explainable Multi-modal Retrieval-augmented Generation
MMRAG-RFT: 两阶段强化微调用于可解释的多模态检索增强生成
专题命中 检索器与排序 :retrieval-augmented generation(title,abstract);分类 cs.AI
AI总结 MMRAG-RFT通过两阶段强化微调提升多模态检索增强生成的可解释性,实现更清晰的推理逻辑和更优的生成效果。
Comments This paper was accepted to AAAI2026