DL4NLP is a Python package inspired by a course at Stanford University by Richard Socher.
The recommended environment is Anaconda with Python 3.4 and PyCharm for development.
For matrix computation, it uses only numpy i.e. you don't need Theano or other framework.
DL4NLP implements word2vec model (skip-gram) proposed by Thomas Mikolov at 2013.
https://code.google.com/archive/p/word2vec/
In order to train word vectors, use this command:
./bin/word2vec.py input.txt output.txt <vector_size> <context_size>
Run the following command on the top of project directory.
python -m unittest discover tests