Are Pre-Trained Convolutions Better Than Pre-Trained Transformers? (2021)arxiv.org2 points·fzliu··0 commentsOpen articleSaveView on HN