April 2022 Summaries

The Vespa Newsletter from April 2022 highlights several updates and improvements to the Vespa platform. Featured in a Vector Podcast episode, Vespa team member Jo Kristian Bergum discusses the evolution of Vespa from sparse to neural search, touching on vector search and multi-stage ranking. The platform now supports a compact tensor format for more efficient data parsing, and introduces modular rank profiles allowing for multiple inheritance and separate file definitions. Enhancements in grouping result capabilities help manage memory usage during data aggregation, while the pyvespa Python bindings have been updated to facilitate deployment to Docker. New query guides aim to aid users in optimizing performance for nearest neighbor searches with multithreaded queries and advanced configurations. The announcement also notes the upcoming release of Vespa 8, which will include breaking changes, and advises users to review release notes to ensure compatibility.