Exploiting Redundancy in Pre-trained Language Models for Efficient Transfer Learning

Publication
Arxiv