Rank records by similarity to the record id. Returns the same shape as
search(): total counts every similar record that passed the filters and
minScore, before the page is cut to limit. Throws when the corpus carries no
frozen embeddings, rather than silently degrading to keyword search — a caller
asking for semantic results should learn that it did not get them.
Rank records by similarity to the record
id. Returns the same shape assearch():totalcounts every similar record that passed the filters andminScore, before the page is cut tolimit. Throws when the corpus carries no frozen embeddings, rather than silently degrading to keyword search — a caller asking for semantic results should learn that it did not get them.