PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier Paper • 2506.10406 • Published Jun 12, 2025 • 2
