benwtrent opened a new issue, #14233:
URL: https://github.com/apache/lucene/issues/14233
### Description
After a6a96cde1c6 Bugfix/fix hnsw search termination check (#14215) HNSW
format recall tests started failing. Need to investigate.
```
TestLucene94HnswVectorsFormat > testRecall FAILED
java.lang.AssertionError: Average recall for DOT_PRODUCT should be at
least 40.0 / 80, got 37
at
__randomizedtesting.SeedInfo.seed([F3F55E436AC90BCA:3E42FDA3F038909D]:0)
at [email protected]/org.junit.Assert.fail(Assert.java:89)
at [email protected]/org.junit.Assert.assertTrue(Assert.java:42)
at
[email protected]/org.apache.lucene.tests.index.BaseKnnVectorsFormatTestCase.assertRecall(BaseKnnVectorsFormatTestCase.java:2009)
at
[email protected]/org.apache.lucene.tests.index.BaseKnnVectorsFormatTestCase.testRecall(BaseKnnVectorsFormatTestCase.java:1940)
at
java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:103)
at java.base/java.lang.reflect.Method.invoke(Method.java:580)
at
[email protected]/com.carrotsearch.randomizedtesting.RandomizedRunner.invoke(RandomizedRunner.java:1758)
at
[email protected]/com.carrotsearch.randomizedtesting.RandomizedRunner$8.evaluate(RandomizedRunner.java:946)
at
[email protected]/com.carrotsearch.randomizedtesting.RandomizedRunner$9.evaluate(RandomizedRunner.java:982)
at
[email protected]/com.carrotsearch.randomizedtesting.RandomizedRunner$10.evaluate(RandomizedRunner.java:996)
```
This makes me think that we have many duplicate vectors in our test graph.
### Gradle command to reproduce
```
./gradlew :lucene:backward-codecs:test --tests
"org.apache.lucene.backward_codecs.lucene94.TestLucene94HnswVectorsFormat.testRecall"
-Ptests.jvms=12 -Ptests.jvmargs= -Ptests.seed=F3F55E436AC90BCA
-Ptests.useSecurityManager=true -Ptests.gui=true -Ptests.file.encoding=UTF-8
-Ptests.vectorsize=128 -Ptests.forceintegervectors=true
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
