
You are building a language model application that must answer questions using information that changes over time. Some teams want to adapt the model with training data, while others want to connect it to an external knowledge base at query time.
What is the difference between fine-tuning and retrieval-augmented generation (RAG)?