An assistant must answer questions about a large document collection that cannot fit in one model request. What is the most appropriate conceptual approach for supplying relevant information within the model context limit?

1. Fine-tune once and assume every future document will automatically be available to the model.
2. Split the text into smaller font sizes before sending it as the same text prompt.
3. Increase temperature until the same model accepts an unlimited number of tokens.
4. Retrieve relevant passages and include a bounded amount of that material with the instructions and user question.