Setup
pip3 install --user beautifulsoup4 sklearn testfixtures unittest2 pyemd morfessor
pip3 install --user pandas==0.20.3
pip3 install --user spacy
pip3 install --user lxml
pip3 install --user numpy
pip3 install --user scipy
pip3 install --user https://github.com/explosion/spacy-models/releases/download/en_core_web_md-1.2.1/en_core_web_md-1.2.1.tar.gz
python3 -m spacy link en_core_web_md en_default
pip3 install --user nltk
printf 'import nltk; nltk.download("wordnet")' | python3
pip3 install --user tensorflow-gpu
pip3 install --user docopt
Requirements
Using slurm version slurm 17.02.2, we ran our experiments using cuda80
and Python version 3.5.2.
git checkout to correct commit
Please run git checkout 50b898e
Downloading resources
Please run bash install.sh
Input data LSTM
Please run bash convert_all_sense_annotations.sh
Running experiments
Please run bash all_lp_jobs.sh
Please run bash evaluate_in_parallel.sh
with the correct arguments.
Please run the bash script to see what these arguments are.
Result tables
Please run python official_results.py
The tables can be found in a folder called paper_tables
.