Re: [PR] SOLR-16667: LTR Add feature vector caching for ranking [solr]

["alessandrobenedetti (via GitHub)" <gi...@apache.org>]

alessandrobenedetti commented on code in PR #3433:
URL: https://github.com/apache/solr/pull/3433#discussion_r2432655283


##########
solr/solr-ref-guide/modules/query-guide/pages/learning-to-rank.adoc:
##########
@@ -131,18 +131,17 @@ See xref:configuration-guide:solr-modules.adoc[Solr Module] for more details.
 <queryParser name="ltr" class="org.apache.solr.ltr.search.LTRQParserPlugin"/>
 ----
 
-* Configuration of the feature values cache.
+* Configuration of the feature values cache which is used for both reranking and feature logging.
+This needs to be added in the `<query>` section as follows.
 +
 [source,xml]
 ----
-<cache name="QUERY_DOC_FV"
-       class="solr.search.CaffeineCache"
-       size="4096"
-       initialSize="2048"
-       autowarmCount="4096"
-       regenerator="solr.search.NoOpRegenerator" />
+<featureVectorCache class="solr.CaffeineCache" size="4096" initialSize="2048" autowarmCount="0"/>
 ----
 
+[NOTE]
+The `featureVectorCache` holds Lucene Document IDs. Since they are transient, this cache is not auto-warmed.

Review Comment:
   The `featureVectorCache` key is computed using the  Lucene Document ID (necessary for document-level features). 
   Since these IDs are transient, this cache does not support auto-warming.



-- 
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-unsubscribe@solr.apache.org

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


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@solr.apache.org
For additional commands, e-mail: issues-help@solr.apache.org