Abstract page for arXiv paper 2608.24949: Demystifying Reinforcement Learning Post-Training of Language Models