December 21, 2021
The DataRobot v7.0.3 release includes some fixed issues in the DataRobot on-premise platform. See the v7.0.0 release notes for:
Issues fixed in v7.0.3¶
The following issues have been fixed since Enterprise release v7.0.2:
- EP-1062: Fixes a Python 3 incompatibility in the Boto library issue when connecting to AWS Cloudwatch behind an HTTPS proxy.
- EP-1345: Fixes an issue that caused a
KeyErrorduring DB migration.
- EP-2133: Updates Scoring Code build plugins to avoid the inclusion of vulnerable log4j versions.
- PLT-3709: Prevents the
ENABLE_DATA_ENGINE_ON_K8Sfeature flag from appearing under Developer UI feature flags for on-prem.
- PLT-3812: Fixes an issue where delegation would not turn off.
- RAPTOR-5240: Fixes an issue where custom model dependency management failed to parse Python package names containing a period (
- RAPTOR-5955: Fixes an issue in the Custom Model Workshop that created a new minor version with empty model content when changing the base environment. Now it properly retains the content of a previous version.
- RAPTOR-6069: Fixes an issue causing the custom models GitHub integration to fail with a SSL handshake error on verification of self-signed certificates.
Trust & Explainability¶
- TED-2688: Fixes issue where the Bias vs Accuracy page would not load when using a fairness metric that's independent of the model's prediction threshold.
- XAI-4106: Fixes the format of the
shapBaseValuefield in v2 Predictions Public API response to be a scalar float value which corresponds with other APIs and documentation.
- XAI-4238: Fixes an issue when loading certain models with SHAP explainers that were built in DataRobot version 5.2.2, which failed to load after upgrading to version 6.3.
COMPUTE_TIMESERIES_AOT_THRESHOLDcan now be configured from an environment.