When using elasticsearch does metadata need to be stored in a metadata object or can the schema be any arbitrary schema? And if so how can I get LlamaIndex to work with any arbitrary schema?

Hi I wanna use the fusion rerank retriever with elastic search, but when I try to intialize the BM25 retriever I can't because the docstore is empty. I already have data ingested in elastic search. Is this still possible?

If I only ever plan on using pre-computed embeddings, setting Settings.embed_model = None gives me an error about an OpenAI key? Is there a way to set embed_model to None like there is with Settings.llm?

When I call .retrieve() I get a NodeWIthScore() object, but the embedding is always None. How can I return the embedding in the NodeWithScore object as well?

def seems like a LlamaIndex issue, LangChain works without an issue

anyone know when the system_prompt argument was removed from the Ollama integration? and how to add one in the new implementation

it seems like CitationQueryEngine is just a fancy prompt? is that correct?

@kapa.ai When I call .retrieve() I get a NodeWIthScore() object, but the embedding is always None. How can I return the embedding in the NodeWithScore object as well?

@kapa.ai How can I print out the intermediate prompts used in TreeSummarizer?

@kapa.ai Is there a way to query an index using an embedding?