PineconeHynridSearchRetriever not having search_kwargs

[danterran00]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

I am using PineconeHybridSearchRetriever class which _get_relevant_documents hasn't implemented yet search_kwargs as parameter. I've added it manually following a sample but now the question is how to use or pass this parameter from PineconeHybridSearchRetriever object.

def _get_relevant_documents(
self, query: str, *, run_manager: CallbackManagerForRetrieverRun,
search_kwargs: Optional[Dict] = None

    result = self.index.query(
        vector=dense_vec,
        sparse_vector=sparse_vec,
        top_k=self.top_k,
        include_metadata=True,
        namespace=self.namespace,
        **(search_kwargs if search_kwargs is not None else {})
    )

As complement on the top I am using MultiQueryRetriever as following

retriever = MultiQueryRetriever(
    retriever = myPineconeHybridSearchRetriever , 
    llm_chain = llm_chain,
    parser_key = "lines",
    include_original = True,
)

Many thanks!

Error Message and Stack Trace (if applicable)

No response

Description

I am using PineconeHybridSearchRetriever class which _get_relevant_documents hasn't implemented yet search_kwargs as parameter. I've added it manually following a sample but now the question is how to use or pass this parameter from PineconeHybridSearchRetriever object.

def _get_relevant_documents(
self, query: str, *, run_manager: CallbackManagerForRetrieverRun,
search_kwargs: Optional[Dict] = None

    result = self.index.query(
        vector=dense_vec,
        sparse_vector=sparse_vec,
        top_k=self.top_k,
        include_metadata=True,
        namespace=self.namespace,
        **(search_kwargs if search_kwargs is not None else {})
    )

As complement on the top I am using MultiQueryRetriever as following

retriever = MultiQueryRetriever(
    retriever = myPineconeHybridSearchRetriever , 
    llm_chain = llm_chain,
    parser_key = "lines",
    include_original = True,
)

Many thanks!

System Info

python 3.11.4
langchain 0.1.0
langchain-community 0.0.10
langchain-core 0.1.33

[QingDeng]

retriever = MultiQueryRetriever(
retriever = myPineconeHybridSearchRetriever ,
llm_chain = llm_chain,
parser_key = "lines",
include_original = True,
)

like this

retriever = MultiQueryRetriever(
retriever = myPineconeHybridSearchRetriever ,
llm_chain = llm_chain,
parser_key = "lines",
include_original = True,
sparse_vector=sparse_vec,
top_k=self.top_k,
)

[Guangdong Liu]

Let me see

[danterran00]

The solution for me was to modify the PineconeHybridSearch class adding the search_kwargs:

search_kwargs: dict = Field(default_factory=lambda: dict(k=100))
"""Keyword arguments to pass to the vectorstore hybrid search."""

and then adding on the function _get_relevant_documents

result = self.index.query(
    vector=dense_vec,
    sparse_vector=sparse_vec,
    top_k=self.top_k,
    include_metadata=True,
    **(self.search_kwargs if self.search_kwargs is not None else {}),
)

You can also have a custom class but I prefer modify this as temp solution