Leveraging Pre-trained Checkpoints for Sequence Generation Tasks

• Published in: Transactions of the Association for Computational Linguistics Vol. 8; pp. 264 - 280
• Main Authors: Rothe, Sascha, Narayan, Shashi, Severyn, Aliaksei
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.01.2020; MIT Press Journals, The; The MIT Press
• Subjects: Automatic summarization; Language; Linguistics; Machine translation; Multilingualism; Natural language; Natural language processing; Sentences
• ISSN: 2307-387X; 2307-387X
• DOI: 10.1162/tacl_a_00313

Unsupervised pre-training of large neural models has recently revolutionized Natural Language Processing. By warm-starting from the publicly released checkpoints, NLP practitioners have pushed the state-of-the-art on multiple benchmarks while saving significant amounts of compute time. So far the focus has been mainly on the Natural Language Understanding tasks. In this paper, we demonstrate the efficacy of pre-trained checkpoints for Sequence Generation. We developed a Transformer-based sequence-to-sequence model that is compatible with publicly available pre-trained BERT, GPT-2, and RoBERTa checkpoints and conducted an extensive empirical study on the utility of initializing our model, both encoder and decoder, with these checkpoints. Our models result in new state-of-the-art results on Machine Translation, Text Summarization, Sentence Splitting, and Sentence Fusion.