Change8

v3.1.3

📦 mlflowView on GitHub →
2 features🐛 5 fixes🔧 4 symbols

Summary

MLflow 3.1.3 introduces artifact permission handling and OpenAI ChatCompletions parsing support while fixing bugs in deployments, model registry, evaluation, and tracking.

✨ New Features

  • Do not copy file permissions when logging artifacts to local artifact repo (Artifacts / Tracking) (#16642).
  • Add support for OpenAI ChatCompletions parse method (Tracking) (#16493).

🐛 Bug Fixes

  • Propagate MLFLOW_DEPLOYMENT_PREDICT_TIMEOUT to databricks-sdk (Deployments) (#16783).
  • Fix issue with search_registered_models when Databricks UC backend does not support filter_string (Model Registry) (#16766).
  • Databricks GenAI evaluation dataset source now returns a DatasetSource instance instead of a string (Evaluation) (#16712).
  • Correct the position of the added tracking_uri parameter in artifact store implementations (Tracking) (#16653).
  • Various small bug fixes and documentation updates (issues #16786, #16692, #16594, #16475).

Affected Symbols