1 paper
Bhavya Vasudeva, Deqing Fu, Tianyi Zhou +3
Transformers achieve state-of-the-art accuracy and robustness across many tasks, but an understanding of their inductive biases and how those biases differ from other neural networ…