Unsupervised Paraphrasing with Pretrained Language Models
This work addresses the costly data collection issue in paraphrase generation for NLP applications, offering a practical unsupervised solution.
The authors tackled the problem of generating high-quality paraphrases without labeled data by proposing an unsupervised training pipeline for pretrained language models, achieving state-of-the-art performance on QQP and ParaNMT datasets with robustness to domain shifts and cross-language transfer.
Paraphrase generation has benefited extensively from recent progress in the designing of training objectives and model architectures. However, previous explorations have largely focused on supervised methods, which require a large amount of labeled data that is costly to collect. To address this drawback, we adopt a transfer learning approach and propose a training pipeline that enables pre-trained language models to generate high-quality paraphrases in an unsupervised setting. Our recipe consists of task-adaptation, self-supervision, and a novel decoding algorithm named Dynamic Blocking (DB). To enforce a surface form dissimilar from the input, whenever the language model emits a token contained in the source sequence, DB prevents the model from outputting the subsequent source token for the next generation step. We show with automatic and human evaluations that our approach achieves state-of-the-art performance on both the Quora Question Pair (QQP) and the ParaNMT datasets and is robust to domain shift between the two datasets of distinct distributions. We also demonstrate that our model transfers to paraphrasing in other languages without any additional finetuning.