It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning Paper • 2609.00638 • Published Sep 1 • 66