Warning: This document is for an old version of rasa NLU. The latest version is 0.15.1.

Installation

rasa NLU itself doesn’t have any external requirements, but to do something useful with it you need to install & configure a backend. Which backend you want to use is up to you.

Setting up rasa NLU

The recommended way to install rasa NLU is using pip:

pip install rasa_nlu

If you want to use the bleeding edge version use github + setup.py:

git clone git@github.com:RasaHQ/rasa_nlu.git
cd rasa_nlu
pip install -r requirements.txt
python setup.py install

rasa NLU allows you to use components to process your messages. E.g. there is a component for intent classification and there are several different components for entity recognition. The different components have their own requirements. To get you started quickly, this installation guide only installs the basic requirements, you may need to install other dependencies if you want to use certain components. When running rasa NLU it will check if all needed dependencies are installed and tell you which are missing, if any.

Note

If you want to make sure you got all the dependencies installed any component might ever need, and you don’t mind the additional dependencies lying around, you can use

pip install -r dev-requirements.txt

to install all requirements.

Setting up a backend

Most of the processing pipeline you can use with rasa NLU either require MITIE, spaCy or sklearn to be installed.

Option 1 : MITIE

The MITIE backend is all-inclusive, in the sense that it provides both the NLP and the ML parts.

pip install git+https://github.com/mit-nlp/MITIE.git

and then download the MITIE models . The file you need is total_word_feature_extractor.dat. Save this somewhere and in your config.json add 'mitie_file' : '/path/to/total_word_feature_extractor.dat'.

Note

Training MITIE can be quite slow on datasets with more than a few intents. You can try
  • to use the sklearn + MITIE backend instead (which uses sklearn for the training) or
  • you can install our mitie fork which should reduce the training time as well.

Option 2 : spaCy + sklearn

You can also run using these two in combination.

installing spacy just requires (for more information visit the spacy docu):

pip install -U spacy
python -m spacy download en

If you haven’t used numpy/scipy before, it is highly recommended that you install and use Anaconda .

Using Anaconda:

conda install scikit-learn
pip install -U sklearn-crfsuite

Using pip:

pip install -U scikit-learn scipy sklearn-crfsuite

Option 3 : sklearn + MITIE

There is a third backend that combines the advantages of the two previous ones:

  1. the fast and good intent classification from sklearn and
  2. the good entitiy recognition and feature vector creation from MITIE

Especially, if you have a larger number of intents (more than 10), training intent classifiers with MITIE can take very long.

To use this backend you need to follow the instructions for installing both, sklearn and MITIE.