@Logan M , One last question, can we

The score? Like the similarity of the retrieved nodes?

print(node.score for node in response.source_nodes)

I need to get the evaluation score for the generated response for the user input with the relevant source_nodes

The evaluation score -- Need to use one of our evaluators then

So unable to get the score in a single shot which means while getting the response for the user input

so need to score by getting the response from OpenAI for the user input and then need to pass the response, user input and the reference answer with the one of our evaluators right?

Yea.

I guess you could use structured outputs and define a response object that includes some evaluation of the response, to get a single shot

https://docs.llamaindex.ai/en/stable/optimizing/advanced_retrieval/structured_outputs/query_engine.html#query-engines-pydantic-outputs