Ensemble Retriever giving me a float error

[Keith-Ho]

Code:
self.vector_store = AzureSearch(azure_search_endpoint=self.endpoint, azure_search_key=self.key, index_name=self.index_name, embedding_function=self.embedding_function, semantic_configuration_name="my-semantic-config)

retriever = self.vector_store.as_retriever(search_type="semantic_hybrid", k=k, semantic_configuration_name="my-semantic-config")

retriever = EnsembleRetriever(retrievers=[retriever])

qa_prompt = ChatPromptTemplate.from_messages([("system", self.prompt), MessagesPlaceHolder("chat_history), ("human", "{input}")])
question_answer_chain = create_stuff_documents_chain(self.llm, qa_prompt)
rag_chain = create_retrieval_chain(retriever, question_answer_chain)
result = rag_chain.invoke({"input"}: question, "chat_history":chat_history

Error:
float() argument must be a string or a real number, not 'NoneType'

Description:
I'm having trouble understanding where this error is coming from. When I don't use the EnsembleRetriever, I have no issues with retrieving an answer from the result. However, when I add it into an EnsembleRetriever, it gives me this error. I have to use the EnsembleRetriever because I'm combining more than one index.

[dosubot[bot]]

Hey @Keith-Ho! 👋 I'm here to help you with bug fixes, answer questions, and guide you on your journey to becoming a contributor.

The error you're encountering with the EnsembleRetriever is likely due to the weights list not being properly initialized or containing None values. This can happen if the weights are not set correctly when using the EnsembleRetriever. The set_weights method in the EnsembleRetriever class is responsible for setting default weights if none are provided. If the weights list is not set correctly, it could lead to this error when the code attempts to use these weights.

To resolve this issue, ensure that the weights list is correctly initialized and does not contain None values. If you do not provide weights, the set_weights method should automatically assign equal weights to all retrievers. Also, verify that the number of weights matches the number of retrievers, as a mismatch will raise a ValueError ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[Keith-Ho]

@dosu

When I had 3 retrievers inside the EnsembleRetriever, I had this code and I was still getting the same error.

retriever = EnsembleRetriever(retrievers=retrievers, weights=[0.4, 0.4, 0.2])

[dosubot[bot]]

Ensure that the retrievers list contains valid retriever instances and that none of them are None. The error message suggests that a NoneType is being passed where a float is expected, which could happen if one of the retrievers is not properly initialized. Double-check that each retriever in the retrievers list is correctly instantiated and that the list matches the length of the weights list. If the issue persists, consider adding debug statements to verify the contents of the retrievers and weights lists before passing them to the EnsembleRetriever.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[Keith-Ho]

Setting k as 50 solved this for me