Comment on Better RAG Results with Reciprocal Rank Fusion and Hybrid SearchparentComments−siquick2yWhen you’re creating your embedding you can store keywords from the content (using an LLM) in the metadata of each chunk which would positively increase the relevancy of results turned from the retrieval.LlamaIndex does this out of the box.−thefourthchime2yThat's interesting! I didn't know that
Comments
When you’re creating your embedding you can store keywords from the content (using an LLM) in the metadata of each chunk which would positively increase the relevancy of results turned from the retrieval.
LlamaIndex does this out of the box.
That's interesting! I didn't know that