Skip to content
Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques such as online, hashing, allreduce, reductions, learning2search, active, and interactive learning.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.github/ISSUE_TEMPLATE Add bug report issue template (#1693) Jan 11, 2019
.pipelines Add clang-format 7.0.1 to CI image (#1719) Feb 2, 2019
.scripts Add example C# Simulator (#1790) Mar 29, 2019
R Fix missing bracket (#1333) Oct 11, 2017
big_tests . Jan 21, 2017
c_test Fix L2S HookTask in c-library and Python bindings (#1818) Apr 4, 2019
cluster [OSX] Improve support for static linking. (#1807) Mar 26, 2019
cs Add example C# Simulator (#1790) Mar 29, 2019
demo [WIP] addressing python semantic checking warnings from issue #1442, … Apr 9, 2018
doc CMake build definitions (#1624) Nov 5, 2018
explore Fix unused params warnings plus incomplete struct init (done to defau… Jan 16, 2019
java vw java 11 compatibility (#1700) Feb 8, 2019
library Fix L2S HookTask in c-library and Python bindings (#1818) Apr 4, 2019
logo_assets . Jan 21, 2017
python Fix conda_install.sh script (#1843) Apr 12, 2019
rapidjson
reinforcement_learning Remove reinforcement_learning (#1675) Nov 13, 2018
sdl Fix issues detected by Secure Development Life check. (SDL-7.0) (#1356) Nov 27, 2017
singularity install netcat Mar 14, 2018
test Fix "average loss" bug in cb_explore.cc (#1825) Apr 18, 2019
utl shift clang-format to advise (#1744) Feb 5, 2019
vowpalwabbit Fix "average loss" bug in cb_explore.cc (#1825) Apr 18, 2019
.clang-format Add clang-format (#1701) Feb 5, 2019
.editorconfig turn off autostyle for new versions and shift to Allman Jul 19, 2017
.gitattributes Fix file enocding for .rc files (#1734) Jan 30, 2019
.gitignore Update Dockerfile for TravisCI, define pipeline, upgrade Java (#1716) Jan 19, 2019
.travis.yml Run coveralls inside docker (#1847) Apr 18, 2019
AUTHORS
CMakeLists.txt Fix else statement (#1832) Apr 8, 2019
INSTALL updated INSTALL Mar 29, 2017
LICENSE
Makefile CMake build definitions (#1624) Nov 5, 2018
README.Windows.md Create Windows build scripts and update instructions (#1721) Jan 23, 2019
README.md
appveyor.yml Create Windows build scripts and update instructions (#1721) Jan 23, 2019
build-linux.sh Add clang-format (#1701) Feb 5, 2019
deployvw.bat fix typo Jun 12, 2015
libvw.pc.in expand Boost_LIBRARIES for pkg-config libs (#1801) Mar 22, 2019
libvw_c_wrapper.pc.in CMake build definitions (#1624) Nov 5, 2018
version.txt CMake build definitions (#1624) Nov 5, 2018

README.md

Vowpal Wabbit

Build Status Windows Build status

Coverage Status Total Alerts Gitter chat

This is the Vowpal Wabbit fast online learning code. For Windows specific info, look at README.Windows.md

Why Vowpal Wabbit?

Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques such as online, hashing, allreduce, reductions, learning2search, active, and interactive learning. There is a specific focus on reinforcement learning with several contextual bandit algorithms implemented and the online nature lending to the problem well. Vowpal Wabbit is a destination for implementing and maturing state of the art algorithms with performance in mind.

  • Input Format. The input format for the learning algorithm is substantially more flexible than might be expected. Examples can have features consisting of free form text, which is interpreted in a bag-of-words way. There can even be multiple sets of free form text in different namespaces.
  • Speed. The learning algorithm is fast -- similar to the few other online algorithm implementations out there. There are several optimization algorithms available with the baseline being sparse gradient descent (GD) on a loss function.
  • Scalability. This is not the same as fast. Instead, the important characteristic here is that the memory footprint of the program is bounded independent of data. This means the training set is not loaded into main memory before learning starts. In addition, the size of the set of features is bounded independent of the amount of training data using the hashing trick.
  • Feature Interaction. Subsets of features can be internally paired so that the algorithm is linear in the cross-product of the subsets. This is useful for ranking problems. The alternative of explicitly expanding the features before feeding them into the learning algorithm can be both computation and space intensive, depending on how it's handled.

Visit the wiki to learn more.

Getting Started

For the most up to date instructions for getting started on Windows, MacOS or Linux please see the wiki. This includes:

You can’t perform that action at this time.