1 paper · 1 filter
Shivanshu Kumar, Gopalakrishnan Srinivasan
While the transformer architecture has achieved state-of-the-art performance on natural language processing tasks, these models impose substantial memory and computational overhead…