Prompt Engineering
Use prompting when the base model is already capable and you mainly need clearer instructions, structure, or role framing at runtime.
Click for explanation
Fine-Tuning vs RAG
Choose the method that changes the right thing.
Prompting changes instructions, RAG changes available knowledge, and fine-tuning changes learned behavior. This page exists to keep readers from mixing those tools together without a reason.
Retrieval-Augmented Generation
Use RAG when the model needs changing knowledge, domain documents, or up-to-date context without changing the underlying model weights.
Click for explanation
Fine-Tuning
Use fine-tuning when the model needs to learn new patterns of behavior, output style, or workflow discipline that prompting and retrieval cannot reliably enforce.
Click for explanation