vector_search

superduper.backends.base.vector_search

Source code

cosine

cosine(x,
     y)

Parameter	Description
x	numpy.ndarray
y	numpy.ndarray, y should be normalized!

Cosine similarity function for vector search.

dot

dot(x,
     y)

Parameter	Description
x	numpy.ndarray
y	numpy.ndarray

Dot function for vector similarity search.

l2

l2(x,
     y)

Parameter	Description
x	numpy.ndarray
y	numpy.ndarray

L2 function for vector similarity search.

VectorSearchBackend

VectorSearchBackend(self)

Base vector-search backend.

BaseVectorSearcher

BaseVectorSearcher(self,
     uuid: str,
     dimensions: int,
     measure: str)

Parameter	Description
uuid	Unique string identifier of index
dimensions	Dimension of the vector embeddings
measure	measure to assess similarity

Base class for vector searchers.

VectorItem

VectorItem(self,
     id: str,
     vector: numpy.ndarray) -> None

Parameter	Description
id	ID of the vector
vector	Vector of the item

Class for representing a vector in vector search with id and vector.