Passing query from SelfQueryRetriever to ContextualCompressionRetriever

[dominikbenk]

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.retrievers.document_compressors import LLMChainFilter
from langchain.retrievers.self_query.base import SelfQueryRetriever
from langchain.retrievers import ContextualCompressionRetriever


retriever = SelfQueryRetriever.from_llm(
    llm, db, document_content_description, metadata_field_info
)
_filter = LLMChainFilter.from_llm(llm)
compression_retriever = ContextualCompressionRetriever(base_compressor=_filter, base_retriever=retriever)

# db document example
docs = [Document(metadata={"size": "M"}, page_content="blue jeans")]

# prompt example
prompt = "blue jeans with size M"

Description

Hello everyone,
lets assume we have the simple implementation of SelfQueryRetriever and ContextualCompressionRetriever from the code example. This is the query extracted from SelfQueryRetriever and used in similarity search:

{
    "query": "blue jeans",
    "filter": "eq(\“size\", M)"
}

This, on the other hand, is the query used in ContextualCompressionRetriever to check for the relevancy (same as original prompt):

{
  "query": "blue jeans with size M"
}

Is it possible to make the query in ContextualCompressionRetriever the same as in SelfQueryRetriever? i.e.:

{
  "query": "blue jeans"
}

Thank you in advance, any help appreciated!

System Info

platform mac
python 3.11