In your solution you fecth all nodes then you filter, but i wanted to know if it was possible to pre-filter before the retrieve . Imagine i have 10K chunks from different documents, i want to retrieve only chunks with document id.
That's the earliest point at which I've found we can filter in llama index. If you have tons, then ... maybe consider using different indexes/doc stores? Not idea, I know. Others probalby know better.