DEEP-GRPO

Deep Dense Exploration for LLM Reinforcement Learning via Pivot-Driven Resampling https://arxiv.org/pdf/2602.14169