ScalarQuantizationCompression Class

Package:

com.azure.search.documents.indexes.models

Maven Artifact:

com.azure:azure-search-documents:11.8.1

• java.lang.Object
• • com.azure.search.documents.indexes.models.VectorSearchCompression
  • • com.azure.search.documents.indexes.models.ScalarQuantizationCompression

public final class ScalarQuantizationCompression
extends VectorSearchCompression

Contains configuration options specific to the scalar quantization compression method used during indexing and querying.

Constructor Summary

Constructor	Description
ScalarQuantizationCompression(String compressionName)	Creates an instance of ScalarQuantizationCompression class.

Method Summary

Modifier and Type	Method and Description
static ScalarQuantizationCompression	fromJson(JsonReader jsonReader) Reads an instance of ScalarQuantizationCompression from the JsonReader.
VectorSearchCompressionKind	getKind() Get the kind property: The name of the kind of compression method being configured for use with vector search.
ScalarQuantizationParameters	getParameters() Get the parameters property: Contains the parameters specific to Scalar Quantization.
ScalarQuantizationCompression	setDefaultOversampling(Double defaultOversampling) Set the defaultOversampling property: Default oversampling factor.
ScalarQuantizationCompression	setParameters(ScalarQuantizationParameters parameters) Set the parameters property: Contains the parameters specific to Scalar Quantization.
ScalarQuantizationCompression	setRerankWithOriginalVectors(Boolean rerankWithOriginalVectors) Set the rerankWithOriginalVectors property: If set to true, once the ordered set of results calculated using compressed vectors are obtained, they will be reranked again by recalculating the full-precision similarity scores.
ScalarQuantizationCompression	setRescoringOptions(RescoringOptions rescoringOptions) Set the rescoringOptions property: Contains the options for rescoring.
ScalarQuantizationCompression	setTruncationDimension(Integer truncationDimension) Set the truncationDimension property: The number of dimensions to truncate the vectors to.
JsonWriter	toJson(JsonWriter jsonWriter)

Methods inherited from VectorSearchCompression

fromJson getCompressionName getDefaultOversampling getKind getRescoringOptions getTruncationDimension isRerankWithOriginalVectors setDefaultOversampling setRerankWithOriginalVectors setRescoringOptions setTruncationDimension toJson

Methods inherited from java.lang.Object

clone equals finalize getClass hashCode notify notifyAll toString wait wait wait

Constructor Details

ScalarQuantizationCompression

public ScalarQuantizationCompression(String compressionName)

Creates an instance of ScalarQuantizationCompression class.

Parameters:

compressionName - the compressionName value to set.

Method Details

fromJson

public static ScalarQuantizationCompression fromJson(JsonReader jsonReader)

Reads an instance of ScalarQuantizationCompression from the JsonReader.

Parameters:

jsonReader - The JsonReader being read.

Returns:

An instance of ScalarQuantizationCompression if the JsonReader was pointing to an instance of it, or null if it was pointing to JSON null.

Throws:

IOException

- If the deserialized JSON object was missing any required properties.

getKind

public VectorSearchCompressionKind getKind()

Get the kind property: The name of the kind of compression method being configured for use with vector search.

Overrides:

ScalarQuantizationCompression.getKind()

Returns:

the kind value.

getParameters

public ScalarQuantizationParameters getParameters()

Get the parameters property: Contains the parameters specific to Scalar Quantization.

Returns:

the parameters value.

setDefaultOversampling

public ScalarQuantizationCompression setDefaultOversampling(Double defaultOversampling)

Set the defaultOversampling property: Default oversampling factor. Oversampling will internally request more documents (specified by this multiplier) in the initial search. This increases the set of results that will be reranked using recomputed similarity scores from full-precision vectors. Minimum value is 1, meaning no oversampling (1x). This parameter can only be set when rerankWithOriginalVectors is true. Higher values improve recall at the expense of latency. For use with only service version 2024-07-01. If using 2025-09-01 or later, use RescoringOptions.defaultOversampling.

Overrides:

ScalarQuantizationCompression.setDefaultOversampling(Double defaultOversampling)

Parameters:

defaultOversampling

setParameters

public ScalarQuantizationCompression setParameters(ScalarQuantizationParameters parameters)

Set the parameters property: Contains the parameters specific to Scalar Quantization.

Parameters:

parameters - the parameters value to set.

Returns:

the ScalarQuantizationCompression object itself.

setRerankWithOriginalVectors

public ScalarQuantizationCompression setRerankWithOriginalVectors(Boolean rerankWithOriginalVectors)

Set the rerankWithOriginalVectors property: If set to true, once the ordered set of results calculated using compressed vectors are obtained, they will be reranked again by recalculating the full-precision similarity scores. This will improve recall at the expense of latency. For use with only service version 2024-07-01. If using 2025-09-01 or later, use RescoringOptions.rescoringEnabled.

Overrides:

ScalarQuantizationCompression.setRerankWithOriginalVectors(Boolean rerankWithOriginalVectors)

Parameters:

rerankWithOriginalVectors

setRescoringOptions

public ScalarQuantizationCompression setRescoringOptions(RescoringOptions rescoringOptions)

Set the rescoringOptions property: Contains the options for rescoring.

Overrides:

ScalarQuantizationCompression.setRescoringOptions(RescoringOptions rescoringOptions)

Parameters:

rescoringOptions

setTruncationDimension

public ScalarQuantizationCompression setTruncationDimension(Integer truncationDimension)

Set the truncationDimension property: The number of dimensions to truncate the vectors to. Truncating the vectors reduces the size of the vectors and the amount of data that needs to be transferred during search. This can save storage cost and improve search performance at the expense of recall. It should be only used for embeddings trained with Matryoshka Representation Learning (MRL) such as OpenAI text-embedding-3-large (small). The default value is null, which means no truncation.

Overrides:

ScalarQuantizationCompression.setTruncationDimension(Integer truncationDimension)

Parameters:

truncationDimension

toJson

public JsonWriter toJson(JsonWriter jsonWriter)

Overrides:

ScalarQuantizationCompression.toJson(JsonWriter jsonWriter)

Parameters:

jsonWriter

Throws:

IOException