MLflow: An open platform to simplify...
summary
MLflow is an open source Databricks platform (launched June, ~57 contributors) that actually implements tracking, projects, models: a logging API that records parameters, metrics and code; a reproducible project spec with Conda/git; and a model format that loads into Python, Spark or a REST server. Demo uses scikit‑learn elastic‑net on a wine dataset (RMSE ~0.77–0.81), logs runs to the UI, and runs batch inference via scikit-learn → Spark UDF or serves models to Kubernetes/SageMaker. Storage backends include S3/GCS/HDFS/FTP and Java/R clients exist.
AI-written summary · about
chapters (5)
- 0:00The challenges of ML
- 4:20Introduction to MLflow
- 10:02MLflow demo and workflow
- 17:46Project roadmap and conclusion
- 18:54Q&A session
related talks
31mMLflow: Platform for Complete Machine Learning Lifecycle
30mMLflow: an Open Source Tool for Managing the ML Project Lifecycle
37mPlatform for Complete Machine Learning Lifecycle
17mIntroducing mlflow
33mOperationalizing Model Lifecycle with MLFlow
01h 25mIntroduction to MLOps with MLflow
30mEnd-to-End MLOps with MLflow and Kubeflow
57mManaging Machine Learning Lifecycle with MLflow