I have created a RAG pipeline using Langchain components, and llama3.
My use case is providing a PDF and ask questions from the PDF. The RAG pipeline can handle the QA part just fine.
I’m using a dense retriever, and a reranker for my implementation.

I needed to know how to handle user requests like, summarize the PDF, or create questions from the PDF where the whole context of the PDF is needed rather than a retrieved chunk. Is there a way to handle these kind of situations?