Meta's official Llama cookbook with getting-started recipes for inference, fine-tuning, and RAG plus end-to-end use-case examples.