Expedia Group’s Keras 3 migration for lodging rankings
An article on Medium reports a two-thirds cut in P99 inference latency, describing the framework migration as an architecture modernization.
TLDR
An article on Medium describes how Keras 3 helped modernize Expedia Group’s lodging ranking stack. It says treating the framework migration as an architecture modernization cut P99 inference latency—the 99th-percentile time needed to generate a model prediction—by two-thirds.
Combined views
32.5K
1 Source, first seen ago