joosthub / pytorch-nlp-tutorial-nyc2017

Materials for O'Reilly DL 4 NLP tutorial (NYC, June 2017)

pytorch-nlp-tutorial

Day 1

Day 1 Slides

Day 1 Data

1. mkdir day_1/data
2. copy contents of this drive in the data folder day_1/data

https://drive.google.com/open?id=0B1sSP-aCtfuHRFJWTkdUbjFUZDQ

3. Download glove and unpack contents in day_1/data/glove
http://nlp.stanford.edu/data/glove.6B.zip

Day 2

Day 2 Slides

Day 2 Data

1. Trump Tweets
2. Not-pruned Names dataset a. Train b. Test c. Day One Version
3. Stanford NLI dataset
4. Amazon Reviews small train

# install anaconda (if needed)

conda create -n dl4nlp python=3.6
source activate dl4nlp
conda install ipython
conda install jupyter
python -m ipykernel install --user --name dl4nlp

# install pytorch
# visit pytorch.org

# assume we are inside a folder dl4nlp
# note: that if you alternatively download the zip and unzip it to
#   a folder, it will be named something else
git clone https://github.com/joosthub/pytorch-nlp-tutorial.git
cd pytorch-nlp-tutorial

pip install -r requirements.txt

# going back to root folder
cd ..

# install torch text
git clone https://github.com/pytorch/text.git
cd text
python setup.py install