MLflow Docker#

一个管理机器学习生命周期的工具

Docker#

docker run -d --name mlflow -p 5000:5000 -v $(pwd)/mlruns:/mlflow/mlruns ghcr.io/mlflow/mlflow mlflow server --host 0.0.0.0 --backend-store-uri sqlite:////mlflow/mlruns/mlflow.db --default-artifact-root /mlflow/mlruns --serve-artifacts

http://localhost:5000

pip#

pip install mlflow
mlflow server --port 5000

Quickstart#

import mlflow
# Connect to remote MLflow server
mlflow.set_tracking_uri("http://localhost:5000/")
mlflow.set_experiment("MLflow Quickstart")
# Enable autologging for scikit-learn
mlflow.sklearn.autolog()