Multiple Generative Models Ensemble for Knowledge-Driven Proactive Human-Computer Dialogue Agent

Code

• github.com/cui0523/Code6/tree/main/duconv
  mindspore★ 0
• github.com/circlePi/knowledge-driven-dialogue-lic2019
  pytorch★ 0
• github.com/2023-MindSpore-1/ms-code-219/tree/main/duconv
  mindspore★ 0
• github.com/lonePatient/knowledge-driven-dialogue-lic2019-rank5
  pytorch★ 0

Abstract

Multiple sequence to sequence models were used to establish an end-to-end multi-turns proactive dialogue generation agent, with the aid of data augmentation techniques and variant encoder-decoder structure designs. A rank-based ensemble approach was developed for boosting performance. Results indicate that our single model, in average, makes an obvious improvement in the terms of F1-score and BLEU over the baseline by 18.67% on the DuConv dataset. In particular, the ensemble methods further significantly outperform the baseline by 35.85%.

Tasks

Reproductions