StyleDGPT: Stylized Response Generation with Pre-trained Language Models
arXiv:2010.02569
Abstract
Generating responses following a desired style has great potentials to extend applications of open-domain dialogue systems, yet is refrained by lacking of parallel data for training. In this work, we explore the challenging task with pre-trained language models that have brought breakthrough to various natural language tasks. To this end, we introduce a KL loss and a style classifier to the fine-tuning step in order to steer response generation towards the target style in both a word-level and a sentence-level. Comprehensive empirical studies with two public datasets indicate that our model can significantly outperform state-of-the-art methods in terms of both style consistency and contextual coherence.
Findings of EMNLP2020
References in corpus (16)
- Sequence to Sequence Learning with Neural Networks
- Categorical Reparameterization with Gumbel-Softmax
- A Neural Conversational Model
- Topic Aware Neural Response Generation
- TransferTransfo: A Transfer Learning Approach for Neural Network Based Conversational Agents
- Towards a Human-like Open-Domain Chatbot
- Conversational AI: The Science Behind the Alexa Prize
- Personalizing Dialogue Agents: I have a dog, do you have pets too?
- Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory
- DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation
- Low-Resource Knowledge-Grounded Dialogue Generation
- Zero-Resource Knowledge-Grounded Dialogue Generation
- What makes a good conversation? How controllable attributes affect human judgments
- A Pre-training Based Personalized Dialogue Generation Model with Persona-sparse Data
- A Lexical, Syntactic, and Semantic Perspective for Understanding Style in Text
- Neural Response Generation with Meta-Words