Hi community. I have a question around vector searches and trying to make it more efficient.
At the moment i implemented Hybrid Search following Alessandro's post. https://sease.io/2023/12/hybrid-search-with-apache-solr.html I got to the point where it's working decently but would like to optimize somehow different areas. I found the concept Binary Quantization and would like to give it a go to test embedding performance, search performance, search recall and precission, memory footprint, disk space etc. As part of the thinking I found DenseVectorField is actually a field of type "float" or "decimal". That means It will use always 4bytes minimun to save the data. My question here would be. Am i right trying this? Is there any obvious reason why Quantaization won't work on Solr? Regards, Sergio Maroto
