TroveGrid
Back to all tools
MLflow logo

MLflow

Open-source ML lifecycle platform — track experiments, version models, deploy anywhere.

0Open Source
MLOps
Official Website View on GitHub
Share:XLinkedInWhatsApp

Overview

MLflow is an open-source platform for managing the ML lifecycle including experiment tracking, model versioning, deployment, and a registry for production models.

Key Features

  • Experiment tracking with metrics, params, and artifacts
  • Model Registry for versioning and stage transitions
  • MLflow Projects for reproducible ML code
  • Serving models via REST API with mlflow models serve
  • Auto-logging for sklearn, TensorFlow, PyTorch, XGBoost
  • Databricks-hosted MLflow for enterprise use

Real-World Workflows

Track model training experiments

  1. 1Start an MLflow run with mlflow.start_run()
  2. 2Log parameters (learning rate, epochs) with mlflow.log_param()
  3. 3Log metrics per epoch with mlflow.log_metric()
  4. 4Compare runs in the MLflow UI at localhost:5000

Register and promote a model to production

  1. 1Log the trained model with mlflow.sklearn.log_model()
  2. 2Register to Model Registry with a version tag
  3. 3Transition version from Staging to Production via UI or API
  4. 4Load the production model in serving with mlflow.pyfunc.load_model()

Getting Started

pip install mlflow

import mlflow
mlflow.set_experiment('my-experiment')

with mlflow.start_run():
    mlflow.log_param('lr', 0.01)
    mlflow.log_metric('accuracy', 0.95)
    mlflow.sklearn.log_model(model, 'model')

# Launch UI
mlflow ui

Compare Alternatives

See how MLflow stacks up against similar tools.

VSwandbVSkubeflowVSbentomlVSray