no code implementations • INLG (ACL) 2020 • Guanyi Chen, Yinhe Zheng, Yupei Du
Personalised response generation enables generating human-like responses by means of assigning the generator a social identity.
no code implementations • ACL 2022 • Yupei Du, Qi Zheng, Yuanbin Wu, Man Lan, Yan Yang, Meirong Ma
To exemplify the potential applications of our study, we also present two strategies (by adding and removing KB triples) to mitigate gender biases in KB embeddings.
1 code implementation • 10 Oct 2023 • Yupei Du, Albert Gatt, Dong Nguyen
Dataset cartography is a simple yet effective dual-model approach that improves the robustness of fine-tuned PLMs.
1 code implementation • 28 Mar 2023 • Olusanmi Hundogan, Xixi Lu, Yupei Du, Hajo A. Reijers
Current methods to generate counterfactual sequences either do not take the process behavior into account, leading to generating invalid or infeasible counterfactual process instances, or heavily rely on domain knowledge.
1 code implementation • 15 Feb 2023 • Yupei Du, Dong Nguyen
Fine-tuning pre-trained language models on downstream tasks with varying random seeds has been shown to be unstable, especially on small datasets.
1 code implementation • EMNLP 2021 • Yupei Du, Qixiang Fang, Dong Nguyen
In this paper, we assess three types of reliability of word embedding gender bias measures, namely test-retest reliability, inter-rater consistency and internal consistency.
no code implementations • 27 Oct 2020 • Guanyi Chen, Yinhe Zheng, Yupei Du
Personalised response generation enables generating human-like responses by means of assigning the generator a social identity.
no code implementations • IJCNLP 2019 • Yupei Du, Yuanbin Wu, Man Lan
Specifically, we use random walk on word association graph to derive bias scores for a large amount of words.