Question Tokens Deserve More Attention: Enhancing Large Language Models without Training through Step-by-Step Reading and Question Attention Recalibration

LRM
Re-assign community