multi-hop-qa

OThink-SRR1: Search, Refine and Reasoning with Reinforced Learning for Large Language Models

OThink-SRR1 is a framework that enhances LLMs with an iterative Search-Refine-Reason process trained via reinforcement learning. It addresses RAG's challenges by distilling relevant facts from retrieved documents, improving efficiency and accuracy in complex multi-hop QA.