borch 0.0.2

Desupervised AI framework

borch

borch is an artificial intelligence (AI) framework developed by Desupervised. It's designed to be flexible and scalable framework that can solve problems using artificial intelligence and machine learning. Doing so by utilizing a wide ranging toolbox including Bayesian inference, ....

It consists of several sub packages:

• ppl: A universal probabilistic programming language
• infer: An inference package with support for Bayesian inference methods such as Variational Inference (VI), Markov Chain Monte Carlo (MCMC) as well as tools for semi-supervised training and many others.
• utils: various utility functions

Usage

Run make help to see available make targets.

Installation

Virtual environment

When installing borch we normally use virtual environment to manage the Python version dependencies. Two good ones are https://virtualenv.pypa.io/en/stable/ and https://docs.conda.io/en/latest/miniconda.html, look at them and pick one to use and follow their documentation to crate and activate an environment.

NB All installations of python packages should be placed in the correct environment. Installing packages in the global python interpreter can result in unexpected behavior, where global packages may be used in favor of local packages.

Install locally

Once an appropriate conda environment has been created, run

make install

to install a production version of borch with support for a GPU, or

ARCH=cpu make install

for a version that only supports a CPU.

To install in development mode on machine(with no gpu support) run, and all development dependencies.

ARCH=cpu make install-dev

and for GPU support use

make install-dev

Docker

Currently, all borch docker images are based on Ubuntu 16.04. The GPU image is based on an Nvidia Cuda version. Both base images are specified as build arguments which calling docker build.

The GPU image can be built using:

docker build --build-arg BASE="nvidia/cuda:9.1-cudnn7-runtime-ubuntu16.04" --build-arg ARCH=gpu --build-arg D2_AWS_REGION=$D2_AWS_REGION --build-arg D2_AWS_ACCESS_KEY_ID=$D2_AWS_ACCESS_KEY_ID --build-arg D2_AWS_SECRET_KEY_ID=$D2_AWS_SECRET_KEY_ID --build-arg D2_AWS_ENDPOINT=$D2_AWS_ENDPOINT --pull -t borch-gpu . 

And the CPU image using:

docker build --build-arg BASE="ubuntu:18.04" --build-arg ARCH=cpu --build-arg D2_AWS_REGION=$D2_AWS_REGION --build-arg D2_AWS_ACCESS_KEY_ID=$D2_AWS_ACCESS_KEY_ID --build-arg D2_AWS_SECRET_KEY_ID=$D2_AWS_SECRET_KEY_ID --build-arg D2_AWS_ENDPOINT=$D2_AWS_ENDPOINT --pull -t borch-cpu .

The D2_AWS_... envioromental variables are optional, but if not provided functionality that relies on s3 buckets will not work.

Contributing

Please read the contribution guidelines in CONTRIBUTING.md.