Release 1.1.2

This release corrects the previous 1.1.1 as TravisCI wasn't able to do the release.

Release 1.1.1

This hotfix version removes the Kotlin dependency from the shaded jar.

Release 1.0.19

Fix Microsoft LightGBM Provider model-locale sensitivity.

Release 1.1.0

This minor version upgrades Guava dependency to version 25.1.

Note: to use this new version, its also necessary to upgrade the Guava to match the same version.

Release 1.0.18

Release 1.0.18

Release 0.5.15

Upgrade the H2O version to 3.30.0.7.
Improve test coverage to ensure that contains H2O algorithms mandatory parameters.

Release 1.0.17

Updates lightgbmlib:2.3.190 to lightgbmlib:3.0.0. On lightgbmlib:3.0.0 we can use UFT-8 characters.
Improve test coverage to ensure that contains H2O algorithms mandatory parameters.

Release 1.0.16

Deploy lightgbmlib:2.3.190 to the remote repository as a separated artifact

LightGBM

• Add support for train with large datasets.
  Before this patch the train dataset matrix couldn't have more than 2,147,483,646 values (#rows * (#features + label)).
• Add multi-threaded training support.

[Broken] Release 1.0.15

Warning: Do not use this version as it doesn't include lightgbmlib:2.3.190 in openml-lightgbm as uber jar. The problem is raised due to the scope provided that was added. Use instead 1.0.16 or any other version above.
Include lightgbmlib:2.3.190 into openml-lightgbm as uber jar

LightGBM

• Add support for train with large datasets.
  Before this patch the train dataset matrix couldn't have more than 2,147,483,646 values (#rows * (#features + label)).
• Add multi-threaded training support.

[Broken] Release 1.0.14

Warning: Do not use this version as it failed to deploy on the remote repository due to timeout, see ticket OSSRH-51097. Use instead 1.0.15 or any other version above.
Include lightgbmlib:2.3.190 into openml-lightgbm as uber jar

LightGBM

• Add support for train with large datasets.
  Before this patch the train dataset matrix couldn't have more than 2,147,483,646 values (#rows * (#features + label)).
• Add multi-threaded training support.