From 50e90aa6ddc6028ca691c7f301452bd75a352c3a Mon Sep 17 00:00:00 2001
From: Joel Knighton <joel.knighton@datastax.com>
Date: Fri, 29 Sep 2023 12:31:24 -0500
Subject: [PATCH] Adjust PQ clustering parameters based on experimentation

---
 .../io/github/jbellis/jvector/pq/ProductQuantization.java     | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/jvector-base/src/main/java/io/github/jbellis/jvector/pq/ProductQuantization.java b/jvector-base/src/main/java/io/github/jbellis/jvector/pq/ProductQuantization.java
index a9e0a0ab7..ca6ce5463 100644
--- a/jvector-base/src/main/java/io/github/jbellis/jvector/pq/ProductQuantization.java
+++ b/jvector-base/src/main/java/io/github/jbellis/jvector/pq/ProductQuantization.java
@@ -39,8 +39,8 @@
  */
 public class ProductQuantization {
     static final int CLUSTERS = 256; // number of clusters per subspace = one byte's worth
-    private static final int K_MEANS_ITERATIONS = 12;
-    private static final int MAX_PQ_TRAINING_SET_SIZE = 256000;
+    private static final int K_MEANS_ITERATIONS = 6;
+    private static final int MAX_PQ_TRAINING_SET_SIZE = 128000;
 
     final float[][][] codebooks;
     final int M;