Repository navigation
v0.37.0
-
educational materials
- Add encrypted fairness-metrics example using fairlearn-fhe by Bader (@BAder82t) in #2623
- Add README for encrypted fairness-metrics notebook by Bader (@BAder82t) in #2624
- Correct co-occurrences typo in data balance README by Eirik Botten Nicolaysen (@avalyset) in #2631
- Correct erroranalysis Python support documentation by Ilya Matiach (@imatiach-msft) in #2667
-
new features
- Add NumPy 1.x/2.x and Pandas 2 support to erroranalysis by Ilya Matiach (@imatiach-msft) in #2665
-
breaking changes
-
Require Python 3.9 or later for responsibleai 0.37.0 and raiwidgets
0.37.0; Python 3.7 and 3.8 are no longer supported. The independently
released erroranalysis, raiutils, and rai-test-utils packages also require
Python 3.9 or later. -
Require Fairlearn
>=0.8.0,<0.9for raiwidgets Pandas 2 reductions
compatibility; raiwidgets 0.36.0 requiredfairlearn==0.7.0. -
Support NumPy 2 while retaining broad runtime ranges for coherent older
environments. Key direct runtime constraints are below; exact tested
combinations belong in CI and image constraints, not universal package
pins. Other requirements and platform markers remain in package metadata.Dependency responsibleai 0.37.0 raiwidgets 0.37.0 numpy >=1.17.2,<3>=1.17.2,<3pandas >=0.25.1,<3>=0.25.1,<3scipy >=1.4.1,<1.15>=1.4.1,<1.14scikit-learn >=0.22.1,!=1.1,<=1.5.1>=0.22.1,<=1.5.1lightgbm >=2.0.11,<4.6>=2.0.11,<5dice-ml >=0.11,<0.13Not direct econml >=0.15.0,<0.17Not direct statsmodels <0.15.0Not direct interpret-community >=0.31.0,<0.34Not direct ml-wrappers >=0.5.4,<0.7Not direct networkx <4Not direct fairlearn Not direct >=0.8.0,<0.9erroranalysis >=0.6.0>=0.6.0raiutils >=0.4.3>=0.4.3rai-core-flask Not direct >=0.7.6responsibleai Not direct ==0.37.0"Not direct" does not mean unconstrained: raiwidgets depends on the matching
responsibleai version and inherits its dependencies. Installing both uses
the intersection of their constraints, including SciPy>=1.4.1,<1.14,
LightGBM>=2.0.11,<4.6, and the scikit-learn!=1.1exclusion.
Interpret Core and SHAP are selected transitively, without direct exact pins.These are declared ranges, not a guarantee that every cross-product works:
transitive requirements can impose higher effective minimums. DiCE 0.11
requires Pandas <2, while DiCE 0.12 requires Pandas >=2. Keep older Pandas
wheels on NumPy 1 even when their metadata lacks a NumPy upper bound.
Official NumPy-2-compatible Pandas wheels start at 2.2.2; this is not a
universal Pandas minimum for NumPy 1 environments. -
Published responsibleai-vision 0.3.8 requires Pandas <2. The tabular
Pandas range still includes Pandas 1.x and does not conflict with that
cap at the metadata level. This does not establish universal runtime
compatibility or NumPy 2 support for vision. Existing vision images need
a coherent legacy NumPy 1/Pandas 1/DiCE 0.11 stack and image-level constraints.
-
-
bug fixes and tests
- Raise a clear error when outcome filters cannot be computed without a
model or predictions. - Allow non-Spark error analysis to load when an installed PySpark version
is incompatible with NumPy 2. - Support both NumPy 1.x and 2.x across the core Python packages and add
compatibility regression coverage. - Retain broad runtime dependency ranges, including Pandas 1.x, DiCE 0.11,
EconML 0.15, and the older scientific stack. Keep exact compatibility
combinations in CI constraints rather than production package pins by
Ilya Matiach (@imatiach-msft) in #2678 - Prevent dashboard browser tests from waiting on external Fluent UI icon
font downloads before the page load event. - Keep the in-repository text and vision package stacks on NumPy 1.x until
their PyTorch dependency floors can support NumPy 2. Publishing these
updated source packages requires separate releases, which are not part
of this tabular release. - Update forecasting and text/vision test paths for removed Pandas 2 APIs.
-
Responsible AI Dashboard
- Fix the Vision data explorer heading accessibility by tongy-msft (@tongyu-microsoft) in #2578
- Support NumPy 1.x and 2.x by Ilya Matiach (@imatiach-msft) in #2642
-
RAI Vision and Text
- Remove detection extra from torchmetrics install by Ilya Matiach (@imatiach-msft) in #2590
-
Error analysis
- Fix object detection conversion with complex predictions and NumPy >=1.26.0 by Ilya Matiach (@imatiach-msft) in #2594
- Correct metrics, cohort outcomes and regression residuals by Sherman Gaucia Laing (@0-Actual) in #2649
- Fix error analysis metric consistency by Ilya Matiach (@imatiach-msft) in #2652
-
RAIUtils
- Fix conversion boundaries and JSON-safe NumPy scalars by Sherman Gaucia Laing (@0-Actual) in #2648
-
Build and test infrastructure
- Fix builds and update to Python 3.9 as default in tests by Ilya Matiach (@imatiach-msft) in #2593
- Fix dependency compatibility for Node 16 and Python library breaking changes by Ilya Matiach (@imatiach-msft) with @Copilot in #2608
- Fix rai-core-flask CI pip 25.x compatibility and artifact upload conflicts by Ilya Matiach (@imatiach-msft) with @Copilot in #2609
- Update Cypress configuration and test assertions by Ilya Matiach (@imatiach-msft) in #2612
- Fix widget-e2e maplibre-gl CSS compilation by Ilya Matiach (@imatiach-msft) in #2613
- Fix CI pipeline failures across all workflows by Ilya Matiach (@imatiach-msft) in #2620
- Harden notebook CI against external failures by Ilya Matiach (@imatiach-msft) in #2653
- Use latest Ubuntu with separate Python 3.9 compatibility lanes by Ilya Matiach (@imatiach-msft) in #2671
- Fix Python setup for rai-test-utils releases by Ilya Matiach (@imatiach-msft) in #2673
- Fix tabular release npm publish output condition by Ilya Matiach (@imatiach-msft) in #2676
- Raise a clear error when outcome filters cannot be computed without a
-
other
-
Package releases and dependency updates
- Release erroranalysis 0.5.5 by Ilya Matiach (@imatiach-msft) in #2595
- Merge CORS origin restrictions for rai-core-flask by Ilya Matiach (@imatiach-msft) in #2618
- Prepare rai-core-flask 0.7.7 by Ilya Matiach (@imatiach-msft) in #2619
- This separate package version is not yet published to PyPI. The current
raiwidgets dependency floor remains rai-core-flask >=0.7.6, so this
tabular release does not guarantee the CORS hardening. Publishing
0.7.7 and raising that floor require separate release/dependency work.
- This separate package version is not yet published to PyPI. The current
- Release erroranalysis 0.6.0 by Ilya Matiach (@imatiach-msft) in #2666
- Release raiutils 0.4.3 by Ilya Matiach (@imatiach-msft) in #2670
- Release rai-test-utils 0.4.3 by Ilya Matiach (@imatiach-msft) in #2674
- Update tabular packages to released utility versions by Ilya Matiach (@imatiach-msft) in #2675
-
Toolchain and dependency maintenance
- Upgrade Nx to 22.7.7 and Node.js to 22.21.1 by Ilya Matiach (@imatiach-msft) in #2562
- Consolidate Dependabot dependency and GitHub Actions updates by Ilya Matiach (@imatiach-msft) in #2641
- Consolidate Dependabot dependency and GitHub Actions updates by Ilya Matiach (@imatiach-msft) in #2657
- Consolidate action pinning and security dependency updates by Ilya Matiach (@imatiach-msft) in #2660
- Bump the github-actions group with 4 updates by Dependabot (@dependabot)[bot] in #2661
- Bump Highcharts from 10.3.3 to 11.4.9 by Dependabot (@dependabot)[bot] in #2662
- Bump Lodash from 4.17.23 to 4.18.1 by Dependabot (@dependabot)[bot] in #2663
- Bump qs from 6.16.0 to 6.16.2 by Dependabot (@dependabot)[bot] in #2664
- Bump the github-actions group with 4 updates by Dependabot (@dependabot)[bot] in #2668
-
Repository maintenance
- Update localization from JUNO by CSIGS@microsoft.com (@csigs) in #2598
- Add Ilya Matiach (@imatiach-msft) as code owner to all repository paths by Ilya Matiach (@imatiach-msft) with @Copilot in #2611
- Pin GitHub Actions to full-length commit SHAs by Dan Fiedler (@danfiedler-msft) in #2628
-
Full Changelog: v0.36.0...v0.37.0