Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translation.

Mozhdeh Gheini, Xiang Ren, Jonathan May

Browse the full EMNLP paper archive.