Efficient transformer with reinforced position embedding for language models

Yen-Che Hsiao, Abhishek Dutta

In this paper, we propose an efficient transformer architecture that uses reinforced positional embedding to obtain superior performance with half the number of encoder decoder lay…