How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias.

Ruiquan Huang, Yingbin Liang, Jing Yang

Browse the full ICML paper archive.