Universal Transformer
2000년 도입 · 논문 18편에서 사용
The Universal Transformer is a generalization of the Transformer architecture. Universal Transformers combine the parallelizability and global receptive field of feed-forward sequence models like the Transformer with the recurrent inductive bias of RNNs. They also utilise a dynamic per-position halting mechanism.
소개 논문: Universal Transformers
Autoregressive Transformers · Natural Language ProcessingTransformers · Natural Language Processing