com.lancedb.lance.index.vector

Class PQBuildParams

java.lang.Object

com.lancedb.lance.index.vector.PQBuildParams

public class PQBuildParams
extends Object

Train a PQ model for a given column.

This will run k-means clustering on each subvector to determine the centroids that will be used to quantize the subvectors. This step runs against a randomly chosen sample of the data. The sample size is typically quite small and PQ training is relatively fast regardless of dataset scale. As a result, accelerators are not needed here.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	PQBuildParams.Builder

Method Summary

Modifier and Type	Method and Description
int	getKmeansRedos()
int	getMaxIters()
int	getNumBits()
int	getNumSubVectors()
int	getSampleRate()
String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Method Detail

getNumSubVectors

public int getNumSubVectors()

getNumBits

public int getNumBits()

getMaxIters

public int getMaxIters()

getKmeansRedos

public int getKmeansRedos()

getSampleRate

public int getSampleRate()

toString

public String toString()

Overrides:

toString in class Object