Glossary term
Glossary term
Foundations
Something done once rather than continuously. The terms static and offline are synonyms. The following are common uses of static and offline in machine learning:
static model (or offline model) is a model trained once and then used for a while.
static training (or offline training) is the process of training a static model.
static inference (or offline inference) is a process in which a model generates a batch of predictions at a time.
Contrast with dynamic.
Created for this library
A risk modeling team uses static features that change rarely, like account opening date, alongside dynamic features in its scorecard.
An ML platform team uses static graphs for production serving to gain throughput from compiler optimizations.
A retail demand team uses static features like product attributes and dynamic features like recent sales in its forecasting model.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License