swagaf

SWAG dataset. More info is at rowanzellers.com/swag. Here's the bibtex again:

@inproceedings{zellers2018swagaf,
    title={SWAG: A Large-Scale Adversarial Dataset for Grounded Commonsense Inference},
    author={Zellers, Rowan and Bisk, Yonatan and Schwartz, Roy and Choi, Yejin},
    booktitle = "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
    year={2018}
}

Setting this up

you'll need to install a bunch of things. I listed my specs if that helps in requirements.txt. The most important thing probably is that I'm using python 3.6, and pytorch 3.1. The following commands will set up the version of allennlp that I used.

git clone https://github.com/allenai/allennlp.git
cd allennlp    
git checkout 7142962d330ca5a95cade114c26a361c78f2042e
INSTALL_TEST_REQUIREMENTS="true" ./scripts/install_requirements.sh
python setup.py install

Also, for everything in this repo, you'll want to export PYTHONPATH=/home/rowan/code/swagaf

Common use cases

The data is all in data/, baseline and instructions are in swag_baselines. Most people will not need to look at create_swag or raw_data but it's there if you need it!