[GitHub] [lucene] benwtrent commented on issue #12342: Prevent VectorSimilarity.DOT_PRODUCT from returning negative scores

[via GitHub]

benwtrent commented on issue #12342:
URL: https://github.com/apache/lucene/issues/12342#issuecomment-1656718447

   OK, here (unless we have done these incorrectly), is the final Cohere test 
(IMO). This is mixing English and Japanese embeddings (just in case the cohere 
model encodes info for language in magnitude). To mix the language embeddings, 
I appended the embedded rows together, shuffled them and then ran the previous 
testing procedure.
   
   # EN-JA
    Reversed         | Ordered | Random
   
:-------------------------:|:-------------------------:|:-------------------------:
   
!![image](https://github.com/apache/lucene/assets/4357155/68e80b3e-de5e-4744-a8a3-5fdf25c21b44)
   
|![image](https://github.com/apache/lucene/assets/4357155/de604446-9635-4a5b-9993-a4bd95f86ae4)|![image](https://github.com/apache/lucene/assets/4357155/2fc51b8b-f7ef-4594-a889-8dc1a4238389)


-- 
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: issues-unsubscr...@lucene.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org