MemoryQueryEmbedding
One already-computed query vector together with the profile it was computed under.
Carried back out of :meth:MemoryService.search so a later recall round can reuse the
round-0 vector instead of paying for a second embedding call. The profile is part of the
value because a vector is only meaningful against the profile that produced it: reusing a
vector across profiles would silently compare incompatible spaces, so a mismatch is
treated as "reuse unavailable" and the round pays for a fresh embedding.
Attributes
attributequery_vectortuple[float, ...]attributeembedding_profileEmbeddingProfileFunctions
constructor__init__(query_vector, embedding_profile) -> Noneparamquery_vectortuple[float, ...]paramembedding_profileEmbeddingProfileReturns
None
