We are happy to announce the availability of MLflow 1.16.0!

In addition to bug and documentation fixes, MLflow 1.16.0 includes the following features and improvements:

  • Add mlflow.pyspark.ml.autolog() API for autologging of pyspark.ml estimators (#4228, @WeichenXu123)
  • Add mlflow.catboost.log_model, mlflow.catboost.save_model, mlflow.catboost.load_model APIs for CatBoost model persistence (#2417, @harupy)
  • Enable mlflow.pyfunc.spark_udf to use column names from model signature by default (#4236, @Loquats)
  • Add datetime data type for model signatures (#4241, @vperiyasamy)
  • Add mlflow.sklearn.eval_and_log_metrics API that computes and logs metrics for the given scikit-learn model and labeled dataset. (#4218, @alkispoly-db)

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