SMRT Chatbots: Improving Non-Task-Oriented Dialog with Simulated Multiple Reference Training
This work addresses the issue of limited conversational data for non-task-oriented chatbots, offering an incremental improvement over existing methods.
The paper tackled the problem of poor quality and non-diverse responses in non-task-oriented dialog models by applying Simulated Multiple Reference Training (SMRT) with a paraphraser to simulate multiple responses per training prompt, resulting in improvements over a Transformer baseline in human and automatic quality scores and lexical diversity, and outperforming pretraining on automatic quality and lexical diversity without needing related-domain data.
Non-task-oriented dialog models suffer from poor quality and non-diverse responses. To overcome limited conversational data, we apply Simulated Multiple Reference Training (SMRT; Khayrallah et al., 2020), and use a paraphraser to simulate multiple responses per training prompt. We find SMRT improves over a strong Transformer baseline as measured by human and automatic quality scores and lexical diversity. We also find SMRT is comparable to pretraining in human evaluation quality, and outperforms pretraining on automatic quality and lexical diversity, without requiring related-domain dialog data.