Search

Your search keyword '"Hu, Xiaowu"' showing total 545 results
Start Over Author "Hu, Xiaowu"
545 results on '"Hu, Xiaowu"'

2. DialogVED: A Pre-trained Latent Variable Encoder-Decoder Model for Dialog Response Generation

Author

Chen, Wei, Gong, Yeyun, Wang, Song, Yao, Bolun, Qi, Weizhen, Wei, Zhongyu, Hu, Xiaowu, Zhou, Bartuer, Mao, Yi, Chen, Weizhu, Cheng, Biao, and Duan, Nan

Subjects
Computer Science - Computation and Language
Abstract

Dialog response generation in open domain is an important research topic where the main challenge is to generate relevant and diverse responses. In this paper, we propose a new dialog pre-training framework called DialogVED, which introduces continuous latent variables into the enhanced encoder-decoder pre-training framework to increase the relevance and diversity of responses. With the help of a large dialog corpus (Reddit), we pre-train the model using the following 4 tasks adopted in language models (LMs) and variational autoencoders (VAEs): 1) masked language model; 2) response generation; 3) bag-of-words prediction; and 4) KL divergence reduction. We also add additional parameters to model the turn structure in dialogs to improve the performance of the pre-trained model. We conduct experiments on PersonaChat, DailyDialog, and DSTC7-AVSD benchmarks for response generation. Experimental results show that our model achieves the new state-of-the-art results on all these datasets., Comment: 13 pages, 1 figures, 9 tables

Published
2022
Full Text
View/download PDF

11. Contextual Fine-to-Coarse Distillation for Coarse-grained Response Selection in Open-Domain Conversations

Author

Chen, Wei, Gong, Yeyun, Xu, Can, Hu, Huang, Yao, Bolun, Wei, Zhongyu, Fan, Zhihao, Hu, Xiaowu, Zhou, Bartuer, Cheng, Biao, Jiang, Daxin, and Duan, Nan

Subjects
Computer Science - Computation and Language
Abstract

We study the problem of coarse-grained response selection in retrieval-based dialogue systems. The problem is equally important with fine-grained response selection, but is less explored in existing literature. In this paper, we propose a Contextual Fine-to-Coarse (CFC) distilled model for coarse-grained response selection in open-domain conversations. In our CFC model, dense representations of query, candidate response and corresponding context is learned based on the multi-tower architecture, and more expressive knowledge learned from the one-tower architecture (fine-grained) is distilled into the multi-tower architecture (coarse-grained) to enhance the performance of the retriever. To evaluate the performance of our proposed model, we construct two new datasets based on the Reddit comments dump and Twitter corpus. Extensive experimental results on the two datasets show that the proposed methods achieve a significant improvement over all evaluation metrics compared with traditional baseline methods.

Published
2021
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results