Low-level primitives for collapsed Gibbs sampling in python and C++
Python C++ CMake Protocol Buffer Makefile Shell HTML
Pull request Compare This branch is 40 commits ahead of forcedotcom:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
benchmarks
cmake/Modules
conda
derivations
distributions
doc
examples
include/distributions
src
.gitignore
.travis.yml
CMakeLists.txt
LICENSE.txt
MANIFEST.in
Makefile
README.md
requirements.txt
setup.py
stream.log
test_cmake.sh
update_license.py

README.md

Build Status Code Quality Latest Version

Distributions

Distributions provides low-level primitives for collapsed Gibbs sampling in Python and C++ including:

  • special numerical functions,
  • samplers and density functions from a variety of distributions,
  • conjugate component models (e.g., gamma-Poisson, normal-inverse-chi-squared),
  • clustering models (e.g., CRP, Pitman-Yor), and
  • efficient wrappers for mixture models.

Distributions powered a machine-learning-as-a-service for Prior Knowledge Inc., and now powers machine learning infrastructure at Salesforce.com.

Installation

For python-only support (no C++) you can install with pip:

pip install distributions

For help with other builds, see the installation documentation.

Documentation

The official documentation lives at http://distributions.readthedocs.org/.

Branch-specific documentation lives at

Authors (alphabetically)

License

Copyright (c) 2014 Salesforce.com, Inc. All rights reserved.

Licensed under the Revised BSD License. See LICENSE.txt for details.