MLflow

About

Open-source ML lifecycle platform. Experiment tracking, model registry, and deployment management.

• MLflow Tracking for recording parameters, metrics, and artifacts from any ML training run
• MLflow Projects for packaging code with dependencies and entry points in a reusable yaml format
• MLflow Models with standardized format (Flavor concept) for multiple deployment targets
• Model Registry with web UI for managing model versions, stage transitions, and annotations
• Pluggable integration with major ML libraries (TensorFlow, PyTorch, Spark, etc.)