From f895eea96e5feb5736b6afa6d138282113e889a7 Mon Sep 17 00:00:00 2001 From: Jiaming Yuan Date: Mon, 5 Jun 2023 08:28:54 +0800 Subject: [PATCH] footnote. --- doc/tutorials/learning_to_rank.rst | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/doc/tutorials/learning_to_rank.rst b/doc/tutorials/learning_to_rank.rst index 72e2123bbec9..b9883d236ac2 100644 --- a/doc/tutorials/learning_to_rank.rst +++ b/doc/tutorials/learning_to_rank.rst @@ -167,7 +167,9 @@ XGBoost implements distributed learning-to-rank with integration of multiple fra Reproducible Result ******************* -Like any other tasks, XGBoost should generate reproducible results given the same hardware and software environments (and data partitions, if distributed interface is used). Even when the underlying environment has changed, the result should still be consistent. However, when the ``lambdarank_pair_method`` is set to ``mean``, XGBoost uses random sampling, and results may differ depending on the platform used. The random number generator used on Windows (Microsoft Visual C++) is different from the ones used on other platforms like Linux (GCC, Clang), so the output varies significantly between these platforms. +Like any other tasks, XGBoost should generate reproducible results given the same hardware and software environments (and data partitions, if distributed interface is used). Even when the underlying environment has changed, the result should still be consistent. However, when the ``lambdarank_pair_method`` is set to ``mean``, XGBoost uses random sampling, and results may differ depending on the platform used. The random number generator used on Windows (Microsoft Visual C++) is different from the ones used on other platforms like Linux (GCC, Clang) [#f0]_, so the output varies significantly between these platforms. + +.. [#f0] `minstd_rand` implementation is different on MSVC. The implementations from GCC and Thrust produce the same output. ********** References