Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

LUCENE-9592: Loosen equality checks in TestVectorUtil. #2084

Merged
merged 1 commit into from
Nov 20, 2020

Conversation

jtibshirani
Copy link
Member

TestVectorUtil occasionally fails because of floating point errors. This
change slightly increases the epsilon in equality checks -- testing shows that
this will greatly decrease the chance of failure.

Copy link
Contributor

@msokolov msokolov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks! Yes, I think your analysis is good - we see people experimenting with quantizing down to a byte-per-sample while maintaining good near-neighbor matching results. https://ai.googleblog.com/2020/07/announcing-scann-efficient-vector.html, for example, showed some promising results in this vein. Are you able to push the commit yet?

@jtibshirani
Copy link
Member Author

Are you able to push the commit yet?

Not yet, I'm still waiting on account credentials.

TestVectorUtil occasionally fails because of floating point errors. This
change slightly increases the epsilon in equality checks -- testing shows that
this will greatly decrease the chance of failure.
@jtibshirani jtibshirani merged commit 8c7b709 into apache:master Nov 20, 2020
@jtibshirani jtibshirani deleted the test-vector-utils branch November 20, 2020 03:05
ctargett pushed a commit to ctargett/lucene-solr that referenced this pull request Dec 3, 2020
TestVectorUtil occasionally fails because of floating point errors. This
change slightly increases the epsilon in equality checks -- testing shows that
this will greatly decrease the chance of failure.
epugh pushed a commit to epugh/lucene-solr-1 that referenced this pull request Jan 15, 2021
TestVectorUtil occasionally fails because of floating point errors. This
change slightly increases the epsilon in equality checks -- testing shows that
this will greatly decrease the chance of failure.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
2 participants