📊 MLflow Flashcards
📌 What is MLflow?
An open-source platform to manage the ML lifecycle, including experimentation, reproducibility, and deployment.
🧪 What is MLflow Tracking?
It records and queries experiments, parameters, metrics, and artifacts from ML runs.
📦 What is MLflow Projects?
A packaging format for reproducible ML code using Git repos and conda/docker environments.
🏷️ What is an MLflow Run?
A single execution of an ML experiment that logs parameters, code, metrics, and outputs.
🧠 What is MLflow Models?
It manages and packages models for diverse deployment formats like pyfunc, sklearn, and H2O.
🚀 What is MLflow Registry?
A centralized store for model versions, lifecycle states (staging/production), and metadata.
🔄 Can you version models with MLflow?
Yes, using the Model Registry with automatic versioning and tagging of deployments.
⚙️ Can MLflow deploy models?
Yes, MLflow supports model serving using REST APIs, SageMaker, Azure ML, and more.
🔗 What integrations does MLflow offer?
Works with TensorFlow, PyTorch, XGBoost, H2O, Scikit-learn, Databricks, Kubernetes, etc.
🔍 How do you visualize metrics?
Via the MLflow UI or programmatically using APIs and client libraries (Python/CLI).
🧩 What is pyfunc in MLflow?
A standard MLflow model format that supports multiple ML libraries with a common interface.
📁 Where are experiment artifacts stored?
They can be stored locally, on cloud (S3, GCS), Azure Blob, or remote servers using artifact URIs.