Uncovering the Role of Initial Saliency in U-Shaped Attention Bias: Scaling Initial Token Weight for Enhanced Long-Text Processing

Zewen Qiang, Sendong Zhao, Haochun Wang, Bing Qin, Ting Liu

Published: 2025, Last Modified: 06 May 2026CoRR 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Loading