domino-experiment-tracking

What problem does it solve?

This Skill streamlines the process of tracking machine learning experiments within Domino Data Lab, making it easier to manage, compare, and reproduce model training runs.

Core Features & Use Cases

• Experiment Management: Set up and manage MLflow experiments for organized tracking.
• Auto-Logging: Automatically capture parameters, metrics, and models for popular ML frameworks.
• Manual Logging: Log custom metrics, parameters, and artifacts for fine-grained control.
• Model Registry: Register, version, and manage models through different stages (Staging, Production).
• Use Case: When training multiple versions of a classification model, use this Skill to log each run's hyperparameters, accuracy, and saved model artifact, then compare them to find the best performing version and register it for deployment.

Quick Start

Use the domino-experiment-tracking skill to set up a new MLflow experiment named 'customer-churn-modeling' and enable auto-logging for scikit-learn.