We are happy to announce the availability of MLflow 1.11.0!

In addition to bug and documentation fixes, MLflow 1.11.0 includes the following features and improvements:

  • New mlflow.sklearn.autolog() API for automatic logging of metrics, params, and models from scikit-learn model training (#3287, @harupy; #3323, #3358 @dbczumar)
  • Registered model & model version creation APIs now support specifying an initial description (#3271, @sueann)
  • The R mlflow_log_model and mlflow_load_model APIs now support XGBoost models (#3085, @lorenzwalthert)
  • New mlflow.list_run_infos fluent API for listing run metadata (#3183, @trangevi)
  • Added section for visualizing and comparing model schemas to model version and model-version-comparison UIs (#3209, @zhidongqu-db)
  • Enhanced support for using the model registry across Databricks workspaces: support for registering models to a Databricks workspace from outside the workspace (#3119, @sueann), tracking run-lineage of these models (#3128, #3164, @ankitmathur-db; #3187, @harupy), and calling mlflow.<flavor>.load_model against remote Databricks model registries (#3330, @sueann)
  • UI support for setting/deleting registered model and model version tags (#3187, @harupy)
  • UI support for archiving existing staging/production versions of a model when transitioning a new model version to staging/production (#3134, @harupy)

For a comprehensive list of changes, see the release change log, and check out the latest documentation on mlflow.org.