carlosandres12 / mlprague18-nlp

NLP Tutorial Workshop at MLPrague 2018

Deep Learning for Text Processing Workshop at Machine Learning Prague 2018

Instructions for attendees of NLP workshop:

• Please come with a working Python installation that includes:
  • Jupyter notebook server
  • Packages: Tensorflow (1.5.0; GPU not required!), Keras (2.1.5), NLTK (3.2.5)
  • Detailed instructions:
    • Linux, MacOSX:
      • create a Python virtual environment by running the following in terminal:

        pip install virtualenv
        virtualenv --system-site-packages pymlprague
        source pymlprague/bin/activate
        pip install tensorflow==1.5.0 keras==2.1.5 nltk==3.2.5 jupyter

    • MacOSX, Windows:
      • If you don’t have Python installed you can download and install Anaconda
      • You can then create a virtual environment with command line:

        conda create -n pymlprague tensorflow=1.5.0 keras=2.1.5 nltk=3.2.5 jupyter
        source activate pymlprague

    • You can find detailed steps for Keras and Tensorflow under the following links, just in case:
      • https://keras.io/#installation
      • https://www.tensorflow.org/install/
  • (Note that we will assume basic familiarity with Python.)
• Download big data required for the experiments you will be doing:
  • https://nlp.stanford.edu/data/glove.6B.zip
  • http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz
  • python -m nltk.downloader semcor punkt perluniprops

Tutorials:

• Movie Reviews Sentiment Classification: IMDB Sentiment.ipynb
• Capital Letters Recognition: uppercase.ipynb

Presentation Slides:

• Workshop Slideshow