This repository contains the official implementation code of the paper DoubleMix: Simple Interpolation-Based Data Augmentation for Text Classification, accepted at COLING 2022.

1. Check the datasets. Training sets of SNLI and MultiNLI can be found in this link. Place them under the folder dataset/snli and dataset/multinli. We implemented the augmentation methods in DoubleMix using files under src/augment folder.

2. Set up the environment

conda create -n doublemix python==3.8
conda activate doublemix
cd DoubleMix/
pip3 install -r requirements.txt

3. Run DoubleMix

cd src/
CUDA_VISIBLE_DEVICES=0 python3 train.py --dataset [dataset] --aug 1

Please cite our paper if you find our work useful for your research:

@inproceedings{chen2022doublemix,
  title={DoubleMix: Simple Interpolation-Based Data Augmentation for Text Classification},
  author={Chen, Hui and Han, Wei and Yang, Diyi and Poria, Soujanya},
  booktitle={Proceedings of the 29th International Conference on Computational Linguistics},
  pages={4622--4632},
  year={2022}
}

Should you have any questions, feel free to contact chchenhui1996@gmail.com.