[jira] [Commented] (SOLR-15778) Compute time required by features extraction

["Christine Poerschke (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SOLR-15778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17466495#comment-17466495 ] 

Christine Poerschke commented on SOLR-15778:
--------------------------------------------

bq. ... to measure how much time 'feature extraction' is consuming ...

[~diegoceccarelli] may recall more detail than me, but from what I see in the code one way is to measure {{DocTransformer.transform}} calls generally (within Solr) and yes it would be much nicer to consider 'feature extraction' specifically and for the measurements to be available outside Solr also.

> Compute time required by features extraction
> --------------------------------------------
>
>                 Key: SOLR-15778
>                 URL: https://issues.apache.org/jira/browse/SOLR-15778
>             Project: Solr
>          Issue Type: Improvement
>    Affects Versions: 8.10.1
>            Reporter: Anna
>            Priority: Minor
>              Labels: learning-to-rank
>
> Currently, in the debug=timing query parameter, no information is given on the time required to execute features extraction (https://solr.apache.org/guide/8_10/common-query-parameters.html#debug-parameter).
> This is crucial for those who use learning to rank and what to improve the temporal performance of learning to rank queries.
> This can help them in differentiating the time required by features extraction (that is separately done for each document to rerank) and the time required by the model to assign a score to each document.



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

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