Thanks Chris! Great tip that! On Tue, Oct 22, 2024, 11:15 AM Guillaume <gjac...@gmail.com> wrote:
> Excellent ! > Thanks ! > > Le mar. 22 oct. 2024 à 20:10, Chris Hostetter <hossman_luc...@fucit.org> a > écrit : > > > > > : Thanks!. That worked. I also wanted to still keep the deduping > mechanism > > : for everything except the knn vectors. SoI ended up excluding the > > embedding > > : fields (whichc were all named embedding_*) from the de-duping > mechanism. > > : > > : <processor class="solr.UniqFieldsUpdateProcessorFactory"> > > : <str name="fieldRegex">(?!.*embedding).*</str> > > : </processor> > > > > You can save yourself some CPU cycles (evaluating that regex against > every > > incoming field name) by using the 'typeClass' selector -- something > > like... > > > > <processor class="solr.UniqFieldsUpdateProcessorFactory"> > > <lst name="exclude"> > > <str name="typeClass">solr.DenseVectorField</str> > > </lst> > > </processor> > > > > > > > https://solr.apache.org/docs/9_7_0/core/org/apache/solr/update/processor/FieldMutatingUpdateProcessorFactory.html > > > > > > -Hoss > > http://www.lucidworks.com/ > > >
