GuidesRecipesAPI ReferenceChangelogDiscussions
Log In

v1.5 Release Notes

October 2022

Fall is here and with it comes new improvements to Modzy!

New Features & Improvements

  • Added deployment support for models that are built to run on ARM devices. This feature is particularly useful for deploying models to edge devices such as the Raspberry Pi and other small form-factor devices that don't use x86 architecture
  • Made significant improvements to the Python SDK:
    • Enabled End-to-end MLOps support making it possible to package, deploy, send to the edge, and execute nearly any ML model within Python
    • Added ability to process all of the contents of a bucket through any model of your choice in one single command. aws-s3, storage-grid, and azure-blob bucket types are all supported.

Bug Fixes

  • Fixed drift boolean assignment in model importer for commercial models, resulting in more commercial models becoming eligible for prediction drift detection
  • Fixed entitlements bug prohibiting data scientists from setting drift baseline configurations and thresholds on their own models
  • Extended model drift update permissions, adding the ability for team admins and project owners to update baselines and thresholds for their respective team/project.