Fetching the paper…

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