I'm trying to set up a simple vector index that has numerous sub indices that are also simple vector. My graph query "works", but I'm unsure how to increase the similarity_top_k to be greater than the default of 1 on either the top level or secondary indices. Is this something that can be accomplished?