Variational Transformers for Diverse Response Generation

TL;DR

This paper introduces Variational Transformers, a novel sequence model combining Transformer architecture with variational latent variables to enhance diversity and relevance in dialogue response generation.

Contribution

It proposes a new Variational Transformer model that incorporates stochastic latent variables into Transformers for more diverse and contextually relevant dialogue responses.

Findings

Models outperform baselines in response diversity.

Enhanced semantic relevance in generated responses.

Improved human judgment scores.

Abstract

Despite the great promise of Transformers in many sequence modeling tasks (e.g., machine translation), their deterministic nature hinders them from generalizing to high entropy tasks such as dialogue response generation. Previous work proposes to capture the variability of dialogue responses with a recurrent neural network (RNN)-based conditional variational autoencoder (CVAE). However, the autoregressive computation of the RNN limits the training efficiency. Therefore, we propose the Variational Transformer (VT), a variational self-attentive feed-forward sequence model. The VT combines the parallelizability and global receptive field of the Transformer with the variational nature of the CVAE by incorporating stochastic latent variables into Transformers. We explore two types of the VT: 1) modeling the discourse-level diversity with a global latent variable; and 2) augmenting the Transformer decoder with a sequence of fine-grained latent variables. Then, the proposed models are evaluated on three conversational datasets with both automatic metric and human evaluation. The experimental results show that our models improve standard Transformers and other baselines in terms of diversity, semantic relevance, and human judgment.