Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Deterministic GPU histogram. (#5361)

Browse Source 
* Use pre-rounding based method to obtain reproducible floating point
  summation.
* GPU Hist for regression and classification are bit-by-bit reproducible.
* Add doc.
* Switch to thrust reduce for `node_sum_gradient`.
This commit is contained in:
Jiaming Yuan
2020-03-04 15:13:28 +08:00
committed by GitHub
parent 9775da02d9
commit 8d06878bf9
18 changed files with 410 additions and 97 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
doc/parameter.rst
View File

@@ -230,6 +230,20 @@ Parameters for Tree Booster
list is a group of indices of features that are allowed to interact with each other.
See tutorial for more information
Additional parameters for `gpu_hist` tree method
================================================
* ``single_precision_histogram``, [default=``false``]
- Use single precision to build histograms. See document for GPU support for more details.
* ``deterministic_histogram``, [default=``true``]
- Build histogram on GPU deterministically. Histogram building is not deterministic due
to the non-associative aspect of floating point summation. We employ a pre-rounding
routine to mitigate the issue, which may lead to slightly lower accuracy. Set to
``false`` to disable it.
Additional parameters for Dart Booster (``booster=dart``)
=========================================================