/self-training-mcqa

Primary LanguagePython

Unsupervised Multiple Choices Question Answering

This is the original implementation of the following paper.

Chi-Liang Liu, Hung-yi Lee. Unsupervised Deep Learning based Multiple Choices Question Answering: Start Learning from Basic Knowledge

Quick Reproduce

pip install -r requirement.txt

Warning: If you want to use fp16, please install apex from https://github.com/NVIDIA/apex. Default config run with fp16.

Then, you can do

./preprocess.sh         # convert mctest and race format to squad format
./run.sh $task $type    # train and eval

task: mctest or race type: gt, highest-only, mml, hard-em

Note: the extracted QA model results are in data/${task}/prediction*.json If you want to use your own prediction, you can overwrite those files.

Usage

For manual running:

python run_multiple_choice.py --cfg config/common.yaml --loss_cfg
config/${loss_type}d.yaml --data_cfg config/${task}.yaml ...(something you want to overwrite the default config varaible)

Citation

@misc{liu2020unsupervised,
      title={Unsupervised Deep Learning based Multiple Choices Question Answering: Start Learning from Basic Knowledge}, 
      author={Chi-Liang Liu and Hung-yi Lee},
      year={2020},
      eprint={2010.11003},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Contact

For any question, please contact Chi-Liang Liu or post a Github issue.