Cosmos QA: Machine Reading Comprehension with Contextual Commonsense Reasoning (EMNLP'2019)

This repository includes the source code and data for Cosmos QA.

Project Page and Leaderboard

Please refer detailed examples, resources, as well as a leaderboard in our project page: Project Page
To measure the performance of your system or create a submission to our leaderboard, please refer to Leaderboard for detailed instructions and policies.

The training/dev/test datasets for Cosmos QA can be found in data/
The BERT with multiway attention model can be ran and tuned with grid_search.sh. We also released the model file that we used in our paper for BERT multiway attention model Here

Python3, Pytorch1.0, tqdm, boto3, requests

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
.idea		.idea
data		data
README.md		README.md
__init__.py		__init__.py
__main__.py		__main__.py
convert_tf_checkpoint_to_pytorch.py		convert_tf_checkpoint_to_pytorch.py
file_utils.py		file_utils.py
grid_search.sh		grid_search.sh
modeling.py		modeling.py
optimization.py		optimization.py
run_multiway_att.py		run_multiway_att.py
tokenization.py		tokenization.py