This repository has been archived by the owner. It is now read-only.
Project for the Advanced Multimodal Machine Learning course at CMU.
Switch branches/tags
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.
bootstrap
data
external
src
tools
.gitattributes
.gitignore
LICENSE
README.md
install.sh
install_docker.sh

README.md

cmu-ammml-project

Project for the Advanced Multimodal Machine Learning course at CMU.

Setup

Install required libraries

Bootstrapping is currently only supported for Ubuntu 14.04. You can use either Theano or TensorFlow as the backend. It's not possible to have both in a single setup since they require different CUDA/cuDNN versions. To bootstrap, simply specify the backend and run the bootstrap script.

$ BACKEND=theano bootstrap/bootstrap_ubuntu.sh
or
$ BACKEND=tensorflow bootstrap/bootstrap_ubuntu.sh

If running without GPUs, set the CPU_ONLY flag:

$ BACKEND=theano CPU_ONLY=1 bootstrap/bootstrap_ubuntu.sh

Reboot the system after this so drivers are properly loaded.

Install Python packages

Once the required libraries have been installed (either through the bootstrap script or manually), you can install the Python packages to a virtual environment using install.sh:

$ ./install.sh

By default, the virtualenv is created inside ./env. You can change the location using the VENV_DIR variable:

$ VENV_DIR=venv ./install.sh

You're ready to go now. You can access the virtualenv by sourcing its activate script:

$ source <VENV_DIR>/bin/activate

To reinstall dependencies, or install new dependencies, simply run the install script again.