Model Persistence

noun

say it: “MOD-ul per-SIS-tunss” /ˈmɑdəl pərˈsɪstəns/

Saving a trained machine learning model to storage so it can be loaded and used later without training it again.

What is model persistence?

Model persistence is saving a trained model to a file so a program can load it later and predict without training again, such as a web app answering requests. In Python that's usually joblib or pickle. Save the whole pipeline and note the library versions, because a file may not load in another version.

In a sentence

  • “Her notes on model persistence recorded the scikit-learn version the file was saved with.”
  • “Model persistence let the web app load the trained pipeline in a second instead of retraining for ten minutes.”

Level: Journeyman · More AI Basics terms at this level

Updated