ML Experiment Tracking
Track experiments, compare runs, and reproduce ML workflows. Experiment tracking is essential for iterative ML development.
Level: Intermediate · Category: MLOps · Estimated time: 6 hours
Prerequisites
- Machine Learning Basics
- PyTorch Mastery
Lessons
- Why Experiment Tracking? — Reproducibility, comparison, and collaboration.
- MLflow Fundamentals — Runs, parameters, metrics, and artifacts.
- MLflow Model Registry — Versioning, staging, and deployment.
- Weights & Biases — Logging, dashboards, and hyperparameter sweeps.
- Dataset Versioning — DVC, Delta Lake, and data lineage.
- Reproducible Pipelines — Environment capture, Docker, and CI/CD.
Topics covered
mlflow, wandb, experiment-tracking, reproducibility, mlops