MLflow 1.28.0 released!

[Weichen Xu]

Hi all,

We are happy to announce the availability of MLflow 1.28.0 !

MLflow 1.28.0 includes several major features and improvements:

Features:

• [Pipelines] Log the full Pipeline runtime configuration to MLflow Tracking during Pipeline execution (#6359, @jinzhang21)
• [Pipelines] Add pipeline.yaml configurations to specify the Model Registry backend used for model registration (#6284, @sunishsheth2009)
• [Pipelines] Support optionally skipping the transform step of the scikit-learn regression pipeline (#6362, @sunishsheth2009)
• [Pipelines] Add UI links to Runs and Models in Pipeline Step Cards on Databricks (#6294, @dbczumar)
• [Tracking] Introduce mlflow.search_experiments() API for searching experiments by name and by tags (#6333, @WeichenXu123; #6227, #6172, #6154, @harupy)
• [Tracking] Increase the maximum parameter value length supported by File and SQL backends to 500 characters (#6358, @johnyNJ)
• [Tracking] Introduce an --older-than flag to mlflow gc for removing runs based on deletion time (#6354, @Jason-CKY)
• [Tracking] Add MLFLOW_SQLALCHEMYSTORE_POOL_RECYCLE environment variable for recycling SQLAlchemy connections (#6344, @postrational)
• [UI] Display deeply nested runs in the Runs Table on the Experiment Page (#6065, @tospe)
• [UI] Add box plot visualization for metrics to the Compare Runs page (#6308, @ahlag)
• [UI] Display tags on the Compare Runs page (#6164, @CaioCavalcanti)
• [UI] Use scientific notation for axes when viewing metric plots in log scale (#6176, @RajezMariner)
• [UI] Add button to Metrics page for downloading metrics as CSV (#6048, @rafaelvp-db)
• [UI] Include NaN and +/- infinity values in plots on the Metrics page (#6422, @hubertzub-db)
• [Tracking / Model Registry] Introduce environment variables to control retry behavior and timeouts for REST API requests (#5745, @peterdhansen)
• [Tracking / Model Registry] Make MlflowClient importable as mlflow.MlflowClient (#6085, @subramaniam02)
• [Model Registry] Add support for searching registered models and model versions by tags (#6413, #6411, #6320, @WeichenXu123)
• [Model Registry] Add stage parameter to set_model_version_tag() (#6185, @subramaniam02)
• [Model Registry] Add --registry-store-uri flag to mlflow server for specifying the Model Registry backend URI (#6142, @Secbone)
• [Models] Improve performance of Spark Model logging on Databricks (#6282, @bbarnes52)
• [Models] Include Pandas Series names in inferred model schemas (#6361, @RynoXLI)
• [Scoring] Make model_uri optional in mlflow models build-docker to support building generic model serving images (#6302, @harupy)
• [R] Support logging of NA and NaN parameter values (#6263, @nathaneastwood)

Bug fixes and documentation updates:

• [Pipelines] Improve scikit-learn regression pipeline latency by limiting dataset profiling to the first 100 columns (#6297, @sunishsheth2009)
• [Pipelines] Use xdg-open instead of open for viewing Pipeline results on Linux systems (#6326, @strangiato)
• [Pipelines] Fix a bug that skipped Step Card rendering in Jupyter Notebooks (#6378, @apurva-koti)
• [Tracking] Use the 401 HTTP response code in authorization failure REST API responses, instead of 500 (#6106, @balvisio)
• [Tracking] Correctly classify artifacts as files and directories when using Azure Blob Storage (#6237, @nerdinand)
• [Tracking] Fix a bug in the File backend that caused run metadata to be lost in the event of a failed write (#6388, @dbczumar)
• [Tracking] Adjust mlflow.pyspark.ml.autolog() to only log model signatures for supported input / output data types (#6365, @harupy)
• [Tracking] Adjust mlflow.tensorflow.autolog() to log TensorFlow early stopping callback info when log_models=False is specified (#6170, @WeichenXu123)
• [Tracking] Fix signature and input example logging errors in mlflow.sklearn.autolog() for models containing transformers (#6230, @dbczumar)
• [Tracking] Fix a failure in mlflow gc that occurred when removing a run whose artifacts had been previously deleted (#6165, @dbczumar)
• [Tracking] Add missing sqlparse library to MLflow Skinny client, which is required for search support (#6174, @dbczumar)
• [Tracking / Model Registry] Fix an mlflow server bug that rejected parameters and tags with empty string values (#6179, @dbczumar)
• [Model Registry] Fix a failure preventing model version schemas from being downloaded with --serve-arifacts enabled (#6355, @abbas123456)
• [Scoring] Patch the Java Model Server to support MLflow Models logged on recent versions of the Databricks Runtime (#6337, @dbczumar)
• [Scoring] Verify that either the deployment name or endpoint is specified when invoking the mlflow deployments predict CLI (#6323, @dbczumar)
• [Scoring] Properly encode datetime columns when performing batch inference with mlflow.pyfunc.spark_udf() (#6244, @harupy)
• [Projects] Fix an issue where local directory paths were misclassified as Git URIs when running Projects (#6218, @ElefHead)
• [R] Fix metric logging behavior for +/- infinity values (#6271, @nathaneastwood)
• [Docs] Move Python API docs for MlflowClient from mlflow.tracking to mlflow.client (#6405, @dbczumar)
• [Docs] Document that MLflow Pipelines requires Make (#6216, @dbczumar)
• [Docs] Improve documentation for developing and testing MLflow JS changes in CONTRIBUTING.rst (#6330, @ahlag)

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