The Art of Scaling Reinforcement Learning Compute for LLMs https://arxiv.org/abs/2510.13786 https://www.alphaxiv.org/ru/overview/2510.13786v1