3.1.0

Highlights

• New native XGBoost provider (openml-xgboost) backed by xgboost4j 3.4.0.
• Supports model import, real-time scoring, and in-process (non-Spark) training.
• Runs on ARM64 (AWS Graviton / Apple Silicon) and Java 25 — unlike the legacy H2O-XGBoost, which is AMD64-only. Exposed in the UI as "DMLC - XGBoost" (binary classifier).

Lineage

3.X release line for clients on the Java 25 baseline (tracked by master).

Changelog

• feat: Implement new xgboost (c865a0d)

Full changelog: 3.0.0...3.1.0

2.2.0

Highlights

• New native XGBoost provider (openml-xgboost) backed by xgboost4j 3.4.0.
• Supports model import, real-time scoring, and in-process (non-Spark) training.
• Runs on ARM64 (AWS Graviton / Apple Silicon) — unlike the legacy H2O-XGBoost, which is AMD64-only. Exposed in the UI as "DMLC - XGBoost" (binary classifier).

Lineage

2.X hotfix line for clients that cannot yet run Java 25 (Java 8+ baseline, tracked by hf-2.X).

Changelog

• feat: Implement new xgboost (d94cc24)
• ci(hf-2.X): fix arm64 job image — maven:3.9-eclipse-temurin-8 (01386b5)

Full changelog: 2.1.2...2.2.0

Release 3.0.0

What's Changed

• feat: update build to use jdk17 by @jpalmeida416 in #155

Full Changelog: 2.1.2...3.0.0

Release 2.1.2

What's Changed

• chore: bump h2o version to 3.46.0.11 by @jpalmeida416 in #153

Full Changelog: 2.1.1...2.1.2

Release 2.1.1

What's Changed

• fix(lightgbm): ensuring field indexes recomputed after removing target and sample weight for model loading by @jpleitao in #152

Full Changelog: 2.1.0...2.1.1

Release 2.1.0

What's Changed

• Add sample weight to LightGBM by @jpleitao in #147
• Bump maven-javadoc-plugin version by @artpdr in #148
• Use Portal OSSRH Staging API by @artpdr in #149
• fix(lightgbm): ensuring c++ resources created in swig are released after training by @jpleitao in #150
• Give id token permissions to fix codecov step in master by @artpdr in #151

New Contributors

• @jpleitao made their first contribution in #147

Full Changelog: 2.0.1...2.1.0

Release 2.0.1

What's Changed

• Add support for Alpine on LGBM by @Renato2000 in #145

New Contributors

• @Renato2000 made their first contribution in #145

Full Changelog: 2.0.0...2.0.1

Release 2.0.0

What's Changed

• Remove distribution of fat-jars with all dependencies by @TiagoMSSantos in #143

Note

This new major release is potentially a breaking change since the projects using feedzai-openml-java might need to add some dependencies in the project that feedzai-openml-java depends on, or may need to exclude some transitive dependencies from feedzai-openml-java if it is already using a different version.

Full Changelog: 1.6.0...2.0.0

Release 1.6.0

What's Changed

• Fixed pinned FairGBM tag to v0.1.0 by @AndreFCruz in #137
• [fix] make-lightgbm docker build by @AlbertoEAF in #140
• Add support for ARM64 on LGBM by @artpdr in #142

New Contributors

• @artpdr made their first contribution in #142

Full Changelog: 1.5.0...1.6.0

1.5.0

What's Changed

• Reverted change to learning_rate that made it mandatory by @antssilva96 in #135
• Document experimental releases in README by @antssilva96 in #136

Note

This release should be used instead of 1.4.X.

When doing an experimental release for 1.4.1 a problem occurred and an actual release was done with version 1.4.9999.