Centre for Language and Speech Technology


“An easily installable distribution of Natural Language Processing software„

What is LaMachine?

LaMachine is a software distribution of NLP software developed by the Language Machines research group and CLST (Radboud University Nijmegen), as well as TiCC (Tilburg University).

Our software is highly specialised and generally depends on a lot of other software. Installing all this software can be a daunting task, compiling it from scratch even more so. Ideally software is installed through your distribution’s package manager, but we do not always have packages available for all platforms, or they may be out of date. LaMachine ensures you can always use all of our software at the very latest stable versions by bundling them all and offering them in three distinct forms:

LaMachine is suitable for both end-users and developers. It has to be noted, however, that running the latest development versions always comes with the risk of decreased stability due to undiscovered bugs.

Pre-installed software

Optional additional software:

Both the VM image as well as the docker image are based on Arch Linux.

Installation & Usage as Virtual Machine (for Linux, BSD, MacOS X, Windows)

  1. Obtain Vagrant from their site or your package manager.
  2. Obtain VirtualBox from their site or your package manager.
  3. Clone this repository and navigate to the directory in the terminal: $ git clone https://github.com/proycon/LaMachine && cd LaMachine (or download the ZIP manually from github)
  4. Power up the VM: vagrant up (this will download and install everything the first time)
  5. SSH into your VM: vagrant ssh
  6. When done, power down the VM with: vagrant halt (and you can delete it entirely with vagrant destroy)

You may want to adapt Vagrantfile to change the number of CPUs and Memory available to the VM (2 CPUs and 3GB RAM by default).

On most Linux distributions, steps one and two may be combined with a simple command such as sudo apt-get install virtualbox vagrant on Ubuntu, or sudo pacman -Syu virtualbox vagrant on Arch Linux.

Entering your LaMachine Virtual Machine as per step 5 should be password-less, other methods may require a login; use username vagrant and password vagrant. The root password is also vagrant.

Various webservices in the Virtual Machine will be automatically accessible through .

Note that LaMachine by default is running on a 64-bit architecture, if you have a 32-bit host OS and really want to run LaMachine despite likely memory shortage; checkout the 32bit branch after step 3 and before step 4 by issuing the following command: git checkout 32bit.

Make sure to also read our privacy section below.

Installation & Usage with Docker (for Linux only)

  1. Obtain Docker from the Docker site (sudo apt-get install docker.io on Debian/Ubuntu).
  2. Pull the LaMachine image: docker pull proycon/lamachine (or the executable may be named docker.io on Debian/Ubuntu)
  3. Start an interactive prompt to LaMachine: docker run -p 8080:80 -t -i proycon/lamachine, or run stuff: docker run proycon/lamachine <program> (use run -i if the program has an interactive mode; set up a mounted volume to pass file from host OS to docker, see here)

There is no need to clone this git repository at all for this method.

Installation & Usage locally (for Linux/BSD/Mac OS X)

LaMachine can also be used on a Linux/BSD/Mac OS X system without root access (provided a set of prerequisites is available on the system!). This is done through an extension for Python VirtualEnv (using Python 3.3 or later), as we provide a lot of Python bindings anyhow. This offers a local environment, ideal for development, that binds against the software globally available on your system. The virtual environment will be contained under a single directory and contains everything. All sources are pulled from git and compiled for you.

  1. Clone this repository in a temporary location (it will only be needed once) and then navigate to this directory in the terminal: $ cd /tmp && git clone https://github.com/proycon/LaMachine (or download the ZIP manually from github and extract it).
  2. In a terminal, navigate to the directory where you want to install LaMachine, for instance your home directory: $ cd ~. A lamachine/ directory that contains everything will be automatically created in the next step. (Advanced users can also pre-create and activate an existing virtual environment that LaMachine will then reuse.)
  3. Bootstrap the virtual environment by calling: /tmp/LaMachine/virtualenv-bootstrap.sh

Note that you will always have to activate your virtual environment with . lamachine/bin/activate (don’t forget the dot!) if you open a new terminal. This also requires you use bash or zsh. To facilitate activation, we recommend you add an alias alias lm=". /path/to/lamachine/bin/activate" in your ~/.bashrc or ~/.zshrc.

You can add the following optional arguments to virtualenv-bootstrap.sh (and lamachine-update.sh):

Tested to work on:

Partially works on:

Updating & Extra Software

Once you have a LaMachine running in whatever form, just run lamachine-update.sh to update everything again.

The lamachine-update.sh script is also used to install additional optional software, pass the optional software as a parameter:

Note that for the docker version, you can pull a new docker image using docker pull proycon/lamachine instead. If you do use lamachine-update.sh with docker, you most likely will want to docker commit your container afterwards to preserve the update!


Unless you explicitly opt-out, LaMachine send a few details to us regarding your installation of LaMachine whenever you install or update it. This is to help us keep track of its usage and improve it.

The following information is sent:

Your IP address will only be used to identify your country and not used in any other way. No personally identifiable information whatsoever will be included in any reports we generate from this and it will never be used for advertisement purposes.

To opt-out of this behaviour, For the virtualenv-boostrap.sh and lamachine-update.sh scripts, add the parameter private. For the VM method, prior to building the VM, edit Vagrantfile and add the private parameter after bootstrap.sh. Due to the nature of Docker, installation of Docker images are not tracked by us (but may be by Docker itself).

LaMachine downloads software from a number of external sources, depending on the form you choose, which may or may not collect your IP:


LaMachine outputs a VERSION file for each installation or upgrade. The version file contains the exact version numbers of all software installed. You can find this file in either in your virtual environment directory or in the root directory (Vagrant/Docker).

You can use the VERSION file to bootstrap LaMachine with specific versions. For the virtual environment form of LaMachine, add the argument version=/path/to/your/VERSIONfile when running virtualenv-bootstrap.sh. For the Vagrant form, substitute the dummy VERSION file with one of your own and adapt Vagrantfile according to the instructions prior to running vagrant up. For Docker, you’ll have to adapt Dockerfile and build the image locally, or rely on an earlier published build.

This versioning is intended to facilitate scientific reproducibility and deployment of production environments. The caveat to always keep in mind is that the versions you run may be outdated and not have any of the latest improvements/fixes applied.

Note that only our own software and certain Python dependencies are subject to this versioning scheme, generic system packages and libraries will always be at their latest versions.


LaMachine comes with several webservices ready out of the box (source: https://github.com/proycon/clamservices). These are RESTful webservices served using CLAM, but also offer a web-interface for human end-users.

In the Virtual Machine variant of LaMachine, these are running and available out-of-the box. In the docker variant, you will need to explicitly start the services first, this is done using the following command from within the container:

sudo /usr/src/LaMachine/startwebservices.sh

All webservices are then accessible through (ensure that this port is free) from your host system. For Docker you have to run the container with the -p 8080:80 for the port forward to be active.

Webservices are currently available for the following software:

For the LaMachine Virtual Environment, however, you have to start and access each service individually using CLAM’s built-in development server:

For FLAT in the virtual environment, run the following:

Each webservice/webapplication will itself advertise on what port it has been launched and how to access it.

Note that there is no or poor authentication enabled on the webservices, so do not expose them to the outside world!


If you have no need for a VM or a self-contained environment, and you have proper administrative access to the system, then it may be possible to install our software using the proper package manager, provided we have packages available:

The final alternative is obtaining all software sources manually (from github or tarballs) and compiling everything yourself, which can be a tedious endeavour.


If you use the Python virtual environment and come across the error undefined symbol: _PyTraceback_Add upon updating LaMachine. Then some dependencies are still making a reference to the global Python interpreter, which has a newer version than the one in the virtual environment. You can fix this issue by copying the newer global version of the Python interpreter into your virtual environment as follows: cp /usr/bin/python3.4 $VIRTUAL_ENV/bin/python3. Then run lamachine-update.sh again.

On some older version you may run into a syntax error error similar to the following when running lamachine-update.sh: /home/proycon/lamachine/bin/lamachine-update.sh: line 775: syntax error near unexpected token `fi' /home/proycon/lamachine/bin/lamachine-update.sh: line 775: `fi' In this case, simply run lamachine-update.sh again and the problem will correct itself.