Using ContextualCompressionRetriever and filtering for Metadata

Short Course Q&A LangChain: Chat with Your Data
1

Hi,
in lesson 5 of the Langchain: Chat with your data course, they are first showing us basic MMR seach, the SelfQueryRetriever, where we can filter for metadata and later on the ContextualCompressionRetriever.

For MMR I can simply pass a filter dictionary to only search in those Documents that are in page 2. For the SelfQueryRetriever, I pass a metadata_info List of AttributeInfo-Objects, to ensure the query knows how to filter.
Yet, for ContextualCompressionRetriever I can’t seem to find a way to include those filters. I want to be able to, for example, look only on page 2 or a specific document of those files.

In the course, they do it like this:

Basic MMR:

docs = vectordb.similarity_search(
    question,
    k=3,
    filter={"source":"docs/cs229_lectures/MachineLearning-Lecture03.pdf"}
)

SelfQueryRetriever

metadata_field_info = [
    AttributeInfo(
        name="source",
        description="The lecture the chunk is from, should be one of `docs/cs229_lectures/MachineLearning-Lecture01.pdf`, `docs/cs229_lectures/MachineLearning-Lecture02.pdf`, or `docs/cs229_lectures/MachineLearning-Lecture03.pdf`",
        type="string",
    ),
    AttributeInfo(
        name="page",
        description="The page from the lecture",
        type="integer",
    ),
]
document_content_description = "Lecture notes"
llm = OpenAI(temperature=0)
retriever = SelfQueryRetriever.from_llm(
    llm,
    vectordb,
    document_content_description,
    metadata_field_info,
    verbose=True
)

and later like this

ContextualCompressionRetriever:

# Wrap our vectorstore
llm = OpenAI(temperature=0)
compressor = LLMChainExtractor.from_llm(llm)

compression_retriever = ContextualCompressionRetriever(
    base_compressor=compressor,
    base_retriever=vectordb.as_retriever(search_type = "mmr")
)

How do I add metadata_field_info and a filter to the ContextualCompressionRetriever?

Initializing it with metadata_field_info = metadata_field_info doesn’t seem to work.

    compression_retriever = ContextualCompressionRetriever(
        base_compressor=compressor,
        base_retriever=vectordb.as_retriever(
            search_type="mmr", metadata_field_info=metadata_field_info
        ),
    )