RAG: Only return sources from metadata, no context

[johkemper]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_core.runnables import RunnableParallel

def format_documents(docs) -> str:
    return "\n\n".join(doc.page_content for doc in docs)


def create_rag_pipeline_with_sourcing(pdf_documents) -> str:
    """Query PDF documents using a parallel processing approach.

    Args:
        pdf_documents: A list of PDF document objects.

    Returns:
        The answer to the query based on the PDF documents content.
    """
    # Create a vector store from PDF documents
    print(f"Number of documents added: {len(pdf_documents)}")

    # Create a vector store from document chunks
    vector_store = Chroma.from_documents(
        documents=data, embedding=OpenAIEmbeddings(), collection_name="pdf"
    )
    print(f"Number of documents in pdf vectorstore: {vector_store._collection.count()}")
    retriever = vector_store.as_retriever()

    # Custom prompt for PDF documents querying
    template_text = """You are a helpful chatbot answering questions. 
    Use the following pieces of context to answer the question at the end.
    If you don't know the answer, just say that you don't know, don't try to make up an answer.
    
    {context}

    Question: {question} 
    
    """
    custom_prompt = PromptTemplate.from_template(template_text)

    # Setup parallel processing chain for querying
    rag_chain_with_source = RunnableParallel(
        {"context": retriever , "question": RunnablePassthrough()}
    ).assign(
        answer=(
            RunnablePassthrough.assign(
                  context=(lambda x: format_documents(x["context"]))

            )
            | custom_prompt
            | llm
            | StrOutputParser()
        )
    )

    # Execute the query and return the response
    return rag_chain_with_source

Description

I would like to get the 'source', eventually the 'page' from the metadata after the 'answer' prompt from the model, which is contained in "metadata={'page': xx, 'source': 'example_source.pdf'}). In addition, I don't want to get the 'context' information.

Hence, the output of the chain should be:

'question': 'blablabla',
'answer': 'blablabla'
'source': 'example_source.pdf'

I'd be grateful if you could help me out!

System Info

langchain
Colab notebook
python3.10

[alice17]

I have the same problem, even opened a discussion. No solution yet.