vislab2013 / DIGITS

Deep Learning GPU Training System

DIGITS (Deep GPU Training System)

DIGITS is is a webapp for training deep learning models.

Installation

DIGITS is supported on Ubuntu 14.04. DIGITS has successfully run on other Linux variants as well as OSX but at this time, only Ubuntu 14.04 is supported.

Prerequisites

DIGITS has several dependencies.

• CUDA
• cuDNN library
• Caffe – NVIDIA branch (version 0.11.0)
• Python packages
• Graphviz

1. CUDA (Either 6.5 or 7.0)

• Download from the CUDA website and follow the installation instructions.

2. cuDNN (v2 Release Candidate 3 required)

• Download from the cuDNN website and follow the installation instructions.

3. NVIDIA branch of Caffe (NVIDIA version 0.11.0)

Full installation directions are at Caffe. Condensed version is as follows:

Install caffe:

% sudo apt-get install git
% cd $HOME
% git clone --branch v0.11.0 https://github.com/NVIDIA/caffe.git
% cd caffe
% sudo apt-get install libatlas-base-dev libatlas-dev libboost-all-dev libopencv-dev
% sudo apt-get install libprotobuf-dev libgoogle-glog-dev libgflags-dev protobuf-compiler
% sudo apt-get install libhdf5-dev libleveldb-dev liblmdb-dev libsnappy-dev
% sudo apt-get install python-pip gfortran
% cd python
% for req in $(cat requirements.txt); do sudo pip install $req; done

Build caffe:

% cd $HOME/caffe
% cp Makefile.config.example Makefile.config
% make all
% make py
% make test
% make runtest

Set environment variables:

% export CAFFE_HOME=${HOME}/caffe

Install DIGITS

% cd $HOME
% git clone https://github.com/NVIDIA/DIGITS.git digits
% cd digits
% sudo apt-get install graphviz gunicorn
% for req in $(cat requirements.txt); do sudo pip install $req; done

Starting the server

You can run DIGITS in two ways:

1. digits-devserver Starts a development server that listens on port 5000 (but you can change the port if you like - try running it with the --help flag).

2. digits-server Starts a gunicorn app that listens on port 8080. If you have installed the nginx.site to your nginx sites-enabled/ directory, then you can view your app at http://localhost/.

Then, check out the Getting Started page.