Deltalm: Encoder-decoder Pre-training For Language Generation And Translation By Augmenting Pretrained Multilingual Encoders

Ma Shuming, Dong Li, Huang Shaohan, Zhang Dongdong, Muzio Alexandre, Singhal Saksham, Awadalla Hany Hassan, Song Xia, Wei Furu. Arxiv 2021

[Paper] [Code]    
Applications Has Code Reinforcement Learning Tools Training Techniques

While pretrained encoders have achieved success in various natural language understanding (NLU) tasks, there is a gap between these pretrained encoders and natural language generation (NLG). NLG tasks are often based on the encoder-decoder framework, where the pretrained encoders can only benefit part of it. To reduce this gap, we introduce DeltaLM, a pretrained multilingual encoder-decoder model that regards the decoder as the task layer of off-the-shelf pretrained encoders. Specifically, we augment the pretrained multilingual encoder with a decoder and pre-train it in a self-supervised way. To take advantage of both the large-scale monolingual data and bilingual data, we adopt the span corruption and translation span corruption as the pre-training tasks. Experiments show that DeltaLM outperforms various strong baselines on both natural language generation and translation tasks, including machine translation, abstractive text summarization, data-to-text, and question generation. The code and pretrained models are available at \url{https://aka.ms/deltalm}.

Similar Work