A Curriculum Learning Approach to Reinforcement Learning: Leveraging RAG for Multimodal Question Answering

Chenliang Zhang, Lin Wang, Yuanyuan Lu +4

This paper describes the solutions of the Dianping-Trust-Safety team for the META CRAG-MM challenge. The challenge requires building a comprehensive retrieval-augmented generation…