DAGsHub-Walkthrough/MLflow Commands.md

MLflow Tracking

This file contains the commands for the MLflow Tracking part in the DAGsHub Walkthrough session.

Install & import MLflow

!pip install mlflow --quiet
import mlflow

Set environment variables:

os.environ['MLFLOW_TRACKING_USERNAME'] = USER_NAME
os.environ['MLFLOW_TRACKING_PASSWORD'] = getpass('Enter your DAGsHub access token or password: ')

• Set remote tracking server mlflow.set_tracking_uri(<ENTER-URL-HERE>)

Logging commands

• Use Auto-logging - mlflow.tensorflow.autolog()
• Manually log parameters - mlflow.log_param("task",2)
• Manually log metrics - mlflow.log_metric("task",2)

For more information please visit:

• DAGsHub Docs
• MLflow Auto Logging docs