joblib

noun

say it: “JOB-lib” /ˈdʒɑbˌlɪb/

A Python package used to save trained machine learning models to files and load them back.

What is joblib?

joblib is a Python package that scikit-learn uses and recommends for saving models: joblib.dump(model, "model.joblib") writes a fitted model to a file, and joblib.load() reads it back, ready to predict. As with pickle, only load files you trust, with the same library versions.

import joblib
from sklearn.linear_model import LinearRegression
model = LinearRegression().fit([[60], [70], [80]], [20, 40, 60])
joblib.dump(model, "lemonade.joblib")
print(joblib.load("lemonade.joblib").predict([[75]]))   # [50.]

In a sentence

  • “The web app used joblib to load the model when it started.”
  • “joblib saved the fitted pipeline as model.joblib.”

Level: Journeyman · More Python terms at this level

Updated