Transformer-based models consist of interleaved feed-forward blocks - that capture content meaning, and relatively more expensive self-attention blocks - that capture context meaning. In this paper, we explored trade-offs and ordering of the blocks to improve upon the current Transformer architecture and proposed PAR Transformer. It needs 35% lower compute time than Transformer-XL achieved by replacing ~63% of the self-attention blocks with feed-forward blocks, and retains the perplexity on WikiText-103 language modelling benchmark. We further validated our results on text8 and enwiki8 datasets, as well as on the BERT model.
@article{arxiv.2009.04534,
title = {Pay Attention when Required},
author = {Swetha Mandava and Szymon Migacz and Alex Fit Florea},
journal= {arXiv preprint arXiv:2009.04534},
year = {2021}
}