How LLMs Are Trained After Pretraining: SFT, Reward Models, and RL Without the Alphabet Soup HackerNoon
