DIGITS is is a webapp for training deep learning models.
- First, check out the instructions below
- Then, ask questions on our user group
- First, check out the Getting Started page
- Then, ask questions on our user group
- Please let us know by filing a new issue
- Bonus points if you want to contribute by opening a pull request!
DIGITS is only officially supported on Ubuntu 14.04. However, DIGITS has been successfully used on other Linux variants as well as on OSX.
DIGITS has several dependencies.
- CUDA
- cuDNN library
- Caffe – NVIDIA branch (version 0.11.0)
- Python packages
- Graphviz
Download from the CUDA website and follow the installation instructions.
Download from the cuDNN website and follow the installation instructions.
Detailed installation instructions are available on caffe's installation page. 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 libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler libatlas-base-dev
% sudo apt-get install python-dev python-pip gfortran
% cd python
% for req in $(cat requirements.txt); do sudo pip install $req; done
Build caffe:
% cd $HOME/caffe
% cp Makefile.config.example Makefile.config
% make all --jobs=4
% make py
Set environment variables:
% export CAFFE_HOME=${HOME}/caffe
% 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
You can run DIGITS in two ways:
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).
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/.
Check out the Getting Started page.