VectorSimilarityFunction
Vector similarity function; used in search to return top K most similar vectors to a target vector. This is a label describing the method used during indexing and searching of the vectors in order to determine the nearest neighbors.
Entries
Dot product. NOTE: this similarity is intended as an optimized way to perform cosine similarity. In order to use it, all vectors must be normalized, including both document and query vectors. Using dot product with vectors that are not normalized can result in errors or poor search results. Floating point vectors must be normalized to be of unit length, while byte vectors should simply all have the same norm.
Cosine similarity. NOTE: the preferred way to perform cosine similarity is to normalize all vectors to unit length, and instead use VectorSimilarityFunction.DOT_PRODUCT. You should only use this function if you need to preserve the original vectors and cannot normalize them in advance. The similarity score is normalised to assure it is positive.
Maximum inner product. This is like VectorSimilarityFunction.DOT_PRODUCT, but does not require normalization of the inputs. Should be used when the embedding vectors store useful information within the vector magnitude
Properties
Functions
Calculates a similarity score between the two vectors with a specified function. Higher similarity scores correspond to closer vectors. Each (signed) byte represents a vector dimension.
Calculates a similarity score between the two vectors with a specified function. Higher similarity scores correspond to closer vectors.
Returns the enum constant of this type with the specified name. The string must match exactly an identifier used to declare an enum constant in this type. (Extraneous whitespace characters are not permitted.)
Returns an array containing the constants of this enum type, in the order they're declared.