yhlleo / reweighting

Reweighting Responses, EMNLP 2018 (short, oral)

Reweighting

Code for paper: Towards Less Generic Responses in Neural Conversation Models: A Statistical Re-weighting Method, EMNLP 2018 (short, oral). [paper] | [code] | [dataset] | [presentation]

Usage

Run the command:

python reweighting.py \
  --src_file /path/to/source_input \
  --tgt_file /path/to/target_input \
  --src_output /path/to/source_output \
  --tgt_output /path/to/source_output \
  --wt_output /path/to/weight_output

Dataset Format

We have provided two paired samples in the examples folder, including a source.train and target.train.

Dataset Download

Please note that this dialogue corpus is RESTRICTED to non-commercial research and educational purposes.

Reference

If you use this research/dataset, please cite our paper:

@inproceedings{liu2018towards,
  title={Towards Less Generic Responses in Neural Conversation Models: A Statistical Re-weighting Method},
  author={Liu, Yahui and Bi, Wei and Gao, Jun and Liu, Xiaojiang and Yao, Jian and Shi, Shuming},
  booktitle={Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing},
  pages={2769--2774},
  year={2018}
}