alternatively, you can see how many chunks being sent to the LLM are actually relevant to your questions on an average query