Training Millions of Personalized Dialogue Agents
This addresses the need for more engaging personalized dialogue agents for users, though it is incremental as it builds on prior work by scaling up data.
The paper tackles the problem of unengaging dialogue systems by introducing a new dataset with 5 million personas and 700 million persona-based dialogues, showing that training at this scale improves end-to-end system performance and achieves state-of-the-art results on a benchmark task.
Current dialogue systems are not very engaging for users, especially when trained end-to-end without relying on proactive reengaging scripted strategies. Zhang et al. (2018) showed that the engagement level of end-to-end dialogue models increases when conditioning them on text personas providing some personalized back-story to the model. However, the dataset used in Zhang et al. (2018) is synthetic and of limited size as it contains around 1k different personas. In this paper we introduce a new dataset providing 5 million personas and 700 million persona-based dialogues. Our experiments show that, at this scale, training using personas still improves the performance of end-to-end systems. In addition, we show that other tasks benefit from the wide coverage of our dataset by fine-tuning our model on the data from Zhang et al. (2018) and achieving state-of-the-art results.