Deep Fusing Pre-trained Models into Neural Machine Translation

Pre-training and fine-tuning have become the de facto paradigm in many natural language processing (NLP) tasks. However, compared to other NLP tasks, neural machine translation (NMT) aims to generate target language sentences through the contextual representation from the source language counterparts. This characteristic means the optimization objective of NMT is far from that of the universal pre-trained models (PTMs), leading to the standard procedure of pretraining and fine-tuning does not work well in NMT. In this paper, we propose a novel framework to deep fuse the pretrained representation into NMT, fully exploring the potential of PTMs in NMT. Specifically, we directly replace the randomly initialized Transformer encoder with a pre-trained encoder and propose a layer-wise coordination structure to coordinate PTM and NMT decoder learning. Then, we introduce a partitioned multi-task learning method to fine-tune the pretrained parameter, reducing the gap between PTM and NMT by progressively learning the task-specific representation. Experimental results show that our approach achieves considerable improvements on WMT14 En2De, WMT14 En2Fr, and WMT16 Ro2En translation benchmarks and outperforms previous work in both autoregressive and non-autoregressive NMT models.