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,
     identifier: str,
     dimensions: int,
     measure: str,
     component: str = 'VectorIndex')

Parameter	Description
identifier	Unique string identifier of index.
dimensions	Number of dimensions of the vectors.
measure	Measure type of the vectors.
component	Component class name.

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.