Announcement

This will be my (@ksaur) last release as maintainer for Hummingbird, as I am moving on to a different company and job role. It has been a fantastic 5 years as maintainer for Hummingbird! Thanks to @interesaaat and @mshr-h for all of the collaborations over the years!

What's Changed

• Pinning onnxruntime to < 1.18.0 to unblock pipeline by @ksaur in #771
• [CI] Update Apache TVM to v0.16.0 by @mshr-h in #772
• skl2onnx==1.17.0 has Unsupported model version by @ksaur in #777
• Deprecating Python 3.8 by @ksaur in #775
• pinning numpy for older SKL backwards compat tests by @ksaur in #780
• [CI] Update Apache TVM to v0.17.0 by @mshr-h in #786
• mitigating onnx install issue by @ksaur in #789
• [CI] Upgrade ubuntu runner image to ubuntu-24.04 by @mshr-h in #787
• [CI] Upgrade macos runner image by @mshr-h in #785
• Fixing codecov token by @ksaur in #796
• Fix TVM error on #797 by @mshr-h in #798
• [CI] Upgrade pytorch to 2.4.1 by @mshr-h in #784

Full Changelog: v0.4.11...v0.4.12

Contributors

• @interesaaat
• @ksaur
• @mshr-h

What's Changed

• Update apache-tvm to v0.15.0 by @mshr-h in #709
• All loss functions which supports SGDClassifer are added by @giriprasad51 in #758
• Forcing users to set override flag to unzip without hash integrity check by @ksaur in #761
• Polynomial features will support degree from 0 to n by @giriprasad51 in #763
• Steps towards xgb==2.0 support by @giriprasad51 in #764
• Support for scikit-learn ==1.4.0 by @ksaur in #756
• Warnings maintenance by @ksaur in #757

Breaking API change for model save/load

• In #761 users are now required to add an additional flag if they wish to unzip without the checksum hashes. Ex: model.load(location, override_flag=True) . This is to prevent any potential exploit of the underlying pickle module, so that users can verify the output checksum for model integrity, or set the override_flag if the model is from a trusted source.

TODOs

• We are aware that documentation is not being generated correctly (#762) due to a permissions change.
• We are continuing to investigate how to support xgb==2.0.

New Contributors

• @giriprasad51 made their first contribution in #758

Special thanks

Thank you to our newest contributor @giriprasad51 for his 3 PRs, and thank you to @mshr-h for all of the effort on apache-tvm to v0.15.0 and the ongoing support!

Full Changelog: v0.4.10...v0.4.11

Contributors

• @ksaur
• @mshr-h
• @giriprasad51

What's Changed

This maintenance release updates to Python 3.11 (for all but TVM), adds support for the current versions (as of Jan 2024) of LGBM and ONNX related libraries.

We are still working on XGB==2.0 support and TVM==0.12.

• pinning xgb to prior 2.0 by @ksaur in #734
• Update devcontainer.json to match the latest property format by @mshr-h in #742
• adding gather to match OHE ONNX changes by @ksaur in #746
• Removing lgbm pin by @ksaur in #748
• Python 3.11 by @ksaur in #749

Thanks to @mshr-h for the ongoing support!

Full Changelog: v0.4.9...v0.4.10

Contributors

• @ksaur
• @mshr-h

What's Changed

• Upgrade PyTorch to 2.0.0 by @mshr-h in #691
• Removing sklearn-onnx pin now that update is released by @ksaur in #693
• Fix SGDClassifier predict result bug by @hikari234 in #703
• Adding hash to save/load for pickle by @ksaur in #713
• Compatibility with scikit-learn 1.3.0 by @ksaur in #721

New Contributors

• @hikari234 made their first contribution in #703

Full Changelog: v0.4.8...v0.4.9

Contributors

• @ksaur
• @mshr-h
• @ytwei3

What's Changed

New support

• Add support for sklearn.mixture.BayesianGaussianMixture by @mshr-h in #677
• Allow for identity POST_TRANSFORM by @jppgks in #674

Bug fixes

• Fix inhomogeneous shape error on numpy array creation by @mshr-h in #667

Maintenance

• Improvements to the testing and pipeline by @mshr-h in #658, #665, #668, #681 and @ksaur in #685, #686
• Support for scikit-learn 1.2 by @ksaur in #661
• Dependabot fixes by @ksaur in #673, #635

Acks

• Thank you very much to @jppgks for their first contribution in #674
• Thank you to @mshr-h for the exciting new BayesianGaussianMixture feature and ongoing support!

Full Changelog: v0.4.7...v0.4.8

Contributors

• @ksaur
• @mshr-h
• @jppgks

What's Changed

• This patch release fixes a bug in ONNX conversion and allows support of varying batch sizes by @stillmatic in #654
• Fixes deprecations in #655

New Contributors

• Thank you to @stillmatic for catching and fixing this bug (#653) and for the additional maintenance work!

Full Changelog: v0.4.6...v0.4.7

Contributors

• @stillmatic

What's Changed

Features

• Adds Lasso, Ridge and ElasticNet by @fd0r in #625
• Added support for more decision conditions in trees and ONNX conversion by @grafail in #631
• Add support for Tweedie, Poisson and Gamma regressors by @interesaaat in #650

Maintenance and fixes

• Remove pinned version for onnxconverter-common by @interesaaat in #618
• Bump action/cache to v3 by @mshr-h in #619
• Fix deprecation warnings for sklearn/scipy by @mshr-h in #610
• deprecating email due to spammers.... by @ksaur in #621
• updating ubuntu version by @ksaur in #623
• Fix linear models conversion when fit_intercept set to False by @RomanBredehoft in #630
• Corrects some typos by @RomanBredehoft in #628
• allow derived types of DataFrame by @liangfu in #637
• [TVM] Unify load params interface by @liangfu in #640
• Update TVM to 0.10 by @mshr-h in #642
• Update to pytorch 1.13 by @interesaaat in #646
• Update actions/checkout and actions/setup-python by @mshr-h in #647
• update codecov/codecov-action to v3 by @mshr-h in #648

New Contributors

• @fd0r made their first contribution in #625
• @RomanBredehoft made their first contribution in #630
• @grafail made their first contribution in #631
• @liangfu made their first contribution in #637

Special Thanks

• Thank you to @mshr-h for the continued support!!

Full Changelog: v0.4.5...v0.4.6

Contributors

• @liangfu
• @interesaaat
• @ksaur
• @mshr-h
• @fd0r
• @grafail
• @melanciani

What's Changed

• Update _decomposition_implementations.py by @interesaaat in #578
• revise example in readme by @vumichien in #579
• Fix problem with new onnx and protobuf by @interesaaat in #582
• Bump TVM to v0.8 by @mshr-h in #581
• Fix the things broken by SKL==1.1.1 by @ksaur in #588
• Use onnxmltools>=1.6.0,<=1.11.0 instead of onnxmltools>=1.6.0 by @mshr-h in #592
• Deprecating Python3.7; updating sklearn-onnx version by @ksaur in #593
• deprecate torch1.7, push to macOS11 by @ksaur in #594
• Fix doc gen by @ksaur in #596
• Fix broken link by @mshr-h in #599
• Fixing documentation generation bug by @ksaur in #598
• Update Dockerfile by @mshr-h in #601
• Use a Microsoft compliant image for docker by @interesaaat in #602
• testing torch==1.12 by @ksaur in #603
• Use n_features_in_ instead of n_features_ by @mshr-h in #604
• Use python -m pip instead of the pip executable by @mshr-h in #605
• check for Sklearn model NotFitted before conversion by @SangamSwadiK in #607
• Upgrade prophet to v1.1 by @mshr-h in #608
• Add Python 3.10 by @mshr-h in #586
• Bump TVM to v0.9 by @mshr-h in #609
• Pinning onnxconverter-common to avoid dep by @ksaur in #614
• Use pre-installed LLVM for building TVM on macOS-11 by @mshr-h in #615

New Contributors

• @vumichien made their first contribution in #579
• @mshr-h made their first contribution in #581
• @SangamSwadiK made their first contribution in #607

Special thanks

Extra special thanks to @mshr-h for the work maintaining our version dependencies and pipeline!

Full Changelog: v0.4.4...v0.4.5

Contributors

• @interesaaat
• @ksaur
• @mshr-h
• @vumichien
• @SangamSwadiK

This minor release includes bug fixes for performance and external dependency updates.

What's Changed

• Verified compatibility with Torch 1.11 released March10 by @ksaur in #572
• Fix xgboost tests for xgb > 1.6.0 by @interesaaat in #576
• Fix for KernelPCA using GPU by @interesaaat in #575

Full Changelog: v0.4.3...v0.4.4

New Contributors

• Thanks to @carolinemckee for the bug report on KPCA #574

Contributors

• @interesaaat
• @ksaur
• @carolinemckee

This minor release includes bug fixes and external dependency updates.

What's Changed

• Fixed a problem with pandas with the latest xgb models in #562
• Minor changes to tests related to verifying that new versions of torch, scikit-learn, and Python3.9 work in #554, #560
• allow tree_implementation="gemm" with onnx backend by @jfrery in #566

New Contributors

• @jfrery made their first contribution in #566
• @shubh0508 contributed a bug fix in #568

Full Changelog: v0.4.2...v0.4.3

Contributors

• @jfrery
• @shubh0508