A Tailored Pre-Training Model for Task-Oriented Dialog Generation
This work addresses a specific problem for developers of task-oriented conversational systems, offering an incremental improvement over existing methods.
The paper tackles the suboptimal performance of general pre-trained language models in task-oriented dialog generation by proposing PRAL, a tailored pre-training model, which shows better or on par results with state-of-the-art methods on three downstream tasks.
The recent success of large pre-trained language models such as BERT and GPT-2 has suggested the effectiveness of incorporating language priors in downstream dialog generation tasks. However, the performance of pre-trained models on the dialog task is not as optimal as expected. In this paper, we propose a Pre-trained Role Alternating Language model (PRAL), designed specifically for task-oriented conversational systems. We adopted (Wu et al., 2019) that models two speakers separately. We also design several techniques, such as start position randomization, knowledge distillation, and history discount to improve pre-training performance. We introduce a task-oriented dialog pretraining dataset by cleaning 13 existing data sets. We test PRAL on three different downstream tasks. The results show that PRAL performs better or on par with state-of-the-art methods.