Patch Release 4.1.2

This patch release is identical to 4.1.0, save for the following fixes:

• Don't load libtreelite.so with symbols visible globally. This breaks downstream packages that embeds Treelite as a library. Global visibility should only be enabled for special environments such as Conda.

Patch release 4.1.1

This patch release is identical to 4.1.0, save for the following fixes:

• Restore support for Python 3.8
• PyPy compatibility patch
• Load libtreelite.so with symbols visible globally

4.1.0 Release

We are excited to announce Treelite 4.1.

What's Changed

• Don't fetch GTest if it's found by CMake by @trxcllnt in #539
• Fix the GTest target name case by @trxcllnt in #541
• Restore support for old binary XGBoost models by @hcho3 in #547
• Enable configuration for custom libpath by @hcho3 in #543
• Clarify output dimensions of GTIL by @hcho3 in #548
• [Breaking] Set 1st dim of prediction output to be row ID by @hcho3 in #549
• Fix GTIL if input does not have sufficient number of columns by @hcho3 in #550
• Support sparse inputs in GTIL by @hcho3 in #551
• Support multi-class, multi-output RandomForestClassifier by @hcho3 in #552
• Fix categorical data handling for HistGradientBoosting in scikit-learn 1.4.0+ by @hcho3 in #553

New Contributors

• @trxcllnt made their first contribution in #539

Full Changelog: 4.0.0...4.1.0

4.0.0 Release

We are excited to announce Treelite 4.0.

What's new

• Native support for multi-target models
• Complete support for XGBoost 2.0, including vector-leaf trees (multi_strategy="multi_output_tree")
• Greater variety of scikit-learn estimators, including HistGradientBoostingRegressor and HistGradientBoostingClassifier.
• Vector base_scores, to support boosting from the average
• Brand-new model builder API with built-in model validation
• Ability to edit trees using field accessor API
• We now provide a pre-built binary wheel targeting ARM64 (aarch64) architecture.

Bug fixes

• Fix base_scores for sklearn GradientBoostingClassifier by @hcho3 in #536
• Fix model concatenation by @hcho3 in #538
• Don't fetch GTest if it's found by CMake by @trxcllnt in #540

Breaking change

• The compiler (C codegen) module has been removed from Treelite. You may use TL2cgen for equivalent functionalities. See the migration guide.

Treelite 4.0.0 Release Candidate 1

We are excited to announce a Release Candidate for Treelite 4.0.

What's new

• Native support for multi-target models
• Complete support for XGBoost 2.0, including vector-leaf trees (multi_strategy="multi_output_tree")
• Greater variety of scikit-learn estimators, including HistGradientBoostingRegressor and HistGradientBoostingClassifier.
• Vector base_scores, to support boosting from the average
• Brand-new model builder API with built-in model validation
• Ability to edit trees using field accessor API

Breaking change

• The compiler (C codegen) module has been removed from Treelite. You may use TL2cgen for equivalent functionalities. See the migration guide.

How to install and test Release Candidate

• Pip: pip install treelite==4.0.0rc1
• Conda: conda install -c conda-forge/label/treelite_rc treelite=4.0.0rc1

Full Changelog: 3.9.1...4.0.0rc1

Patch release 3.9.1

This release is almost identical to 3.9.0 release. It was updated to be compatible with XGBoost 2.0.

3.9.0 Release

Deprecation notice

The following parts of Treelite are deprecated and will be removed in Treelite 4.0.0.
You may use TL2cgen for equivalent functionalities. See the migration guide.

• treelite.Model.export_lib
• treelite.Model.export_srcpkg
• treelite.Model.compile
• treelite.Annotator
• treelite.generate_makefile
• treelite.generate_cmakelists
• treelite.create_shared
• treelite_runtime.Predictor
• treelite_runtime.DMatrix

What's Changed

• Don't install gtest if building from source by @hcho3 in #491
• Deprecate C codegen by @hcho3 in #493
• [CI] Make nightly builds of CPack for amd64 and aarch64 by @hcho3 in #494

Full Changelog: 3.4.0...3.9.0

3.4.0 Release

What's Changed

• Rename byte serializer by @hcho3 in #482
• Improve CMake build by @hcho3 in #484
• Add GetVersion() to Model object by @hcho3 in #485
• Replace deprecated pkg_resources by @hcho3 in #486
• Replace deprecated function in Python layer by @hcho3 in #487

Full Changelog: 3.3.0...3.4.0

3.3.0 Release

What's Changed

• Add instruction to build sdist in release script by @hcho3 in #462
• [CI] Use Python 3.8 by @hcho3 in #476
• Set type fields in ConcatenateModelObjects() by @hcho3 in #475
• Add API function to serialize to strings by @hcho3 in #480

Full Changelog: 3.2.0...3.3.0

3.2.0 Release

The 3.2 release implements three new features:

• predict_leaf: predict leaf node IDs
• predict_per_tree: obtain prediction per tree
• import_from_json: build Treelite model using a JSON string

What's Changed

• Add allow_unknown_field flag to XGBoost JSON parser by @hcho3 in #423
• Bump development version by @hcho3 in #437
• [CI] Fix test timeout for test_extra_field_in_xgb_json by @hcho3 in #436
• [CI] Require C++17 by @hcho3 in #440
• Implement predict_leaf and predict_per_tree by @hcho3 in #442
• Improve pytest coverage for GTIL by @hcho3 in #447
• Add warning when loading from newer minor version by @hcho3 in #450
• Fix docstring for predict_leaf by @hcho3 in #451
• Add ability to import model from JSON by @hcho3 in #448
• Initial support for HistGradientBoostingClassifier / HistGradientBoostingRegressor by @hcho3 in #444
• Fix up documentation by @hcho3 in #454
• Fix up documentation, round 2 by @hcho3 in #455
• Fix up documentation, round 3 by @hcho3 in #456
• Fix link for Doxygen by @hcho3 in #457
• Add tutorial for JSON import by @hcho3 in #459

Full Changelog: 3.1.0...3.2.0