Dior-CVAE: Pre-trained Language Models and Diffusion Priors for Variational Dialog Generation

Tianyu Yang, Thy Thy Tran, Iryna Gurevych


Abstract
Current variational dialog models have employed pre-trained language models (PLMs) to parameterize the likelihood and posterior distributions. However, the Gaussian assumption made on the prior distribution is incompatible with these distributions, thus restricting the diversity of generated responses. These models also suffer from posterior collapse, i.e., the decoder tends to ignore latent variables and directly access information captured in the encoder through the cross-attention mechanism. In this work, we propose Dior-CVAE, a hierarchical conditional variational autoencoder (CVAE) with diffusion priors to address these challenges. We employ a diffusion model to increase the complexity of the prior distribution and its compatibility with the distributions produced by a PLM. Also, we propose memory dropout to the cross-attention mechanism, which actively encourages the use of latent variables for response generation. Overall, experiments across two commonly used open-domain dialog datasets show that our method can generate more diverse responses without large-scale dialog pre-training. Code is available at https://github.com/UKPLab/dior-cvae.
Anthology ID:
2023.findings-emnlp.313
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2023
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4718–4735
Language:
URL:
https://aclanthology.org/2023.findings-emnlp.313
DOI:
10.18653/v1/2023.findings-emnlp.313
Bibkey:
Cite (ACL):
Tianyu Yang, Thy Thy Tran, and Iryna Gurevych. 2023. Dior-CVAE: Pre-trained Language Models and Diffusion Priors for Variational Dialog Generation. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 4718–4735, Singapore. Association for Computational Linguistics.
Cite (Informal):
Dior-CVAE: Pre-trained Language Models and Diffusion Priors for Variational Dialog Generation (Yang et al., Findings 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.findings-emnlp.313.pdf