How LLMs are trained

Pre-training, fine-tuning, human feedback, scaling and limits.