Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translation
This incremental finding helps researchers and practitioners mitigate catastrophic forgetting and extend models to new language pairs more efficiently.
The paper tackles the problem of adapting pretrained Transformers for machine translation by showing that fine-tuning only cross-attention parameters is nearly as effective as full fine-tuning, enabling cross-lingually aligned embeddings and reduced parameter overhead.
We study the power of cross-attention in the Transformer architecture within the context of transfer learning for machine translation, and extend the findings of studies into cross-attention when training from scratch. We conduct a series of experiments through fine-tuning a translation model on data where either the source or target language has changed. These experiments reveal that fine-tuning only the cross-attention parameters is nearly as effective as fine-tuning all parameters (i.e., the entire translation model). We provide insights into why this is the case and observe that limiting fine-tuning in this manner yields cross-lingually aligned embeddings. The implications of this finding for researchers and practitioners include a mitigation of catastrophic forgetting, the potential for zero-shot translation, and the ability to extend machine translation models to several new language pairs with reduced parameter storage overhead.