Based on the work of Hareesh Bahuleyan et al. we modified the structure of the model so that the model can take into account the correlation in the encode-decode process.
The proposed model and baselines have been evaluated on two experiments:
Neural Question Generation with the SQuAD dataset https://rajpurkar.github.io/SQuAD-explorer/
Daily Dialog dataset http://yanran.li/dailydialog.html
tensorflow-gpu==1.3.0
Keras==2.0.8
numpy==1.12.1
pandas==0.22.0
gensim==3.1.2
nltk==3.4.5
tqdm==4.19.1
- Hareesh Bahuleyan, Lili Mou, Hao Zhou, and Olga Vechtomova, “Stochastic wasserstein autoencoder for probabilistic sentence generation,” in NAACL, 2019, pp. 4068–4076. https://github.com/HareeshBahuleyan/probabilistic_nlg
- Hareesh Bahuleyan, Lili Mou, Olga Vechtomova, and Pascal Poupart, “Variational attention for sequence-to- sequence models,” in COLING, 2018, pp. 1672–1682. https://github.com/HareeshBahuleyan/tf-var-attention
- 深度学习的互信息：无监督提取特征 https://kexue.fm/archives/6024, https://github.com/bojone/infomax
If you find our source useful, please consider citing our work.
Zhang X, Li Y, Peng X, et al. Correlation encoder-decoder model for text generation[C]//2022 International Joint Conference on Neural Networks (IJCNN). IEEE, 2022: 1-7.