TY  - RPRT
TI  - Unveiling and Harnessing Hidden Attention Sinks: Enhancing Large Language Models without Training through Attention Calibration
AU  - Zhongzhi Yu
AU  - Zheng Wang
AU  - Yonggan Fu
AU  - Huihong Shi
AU  - Khalid Shaikh
AU  - Yingyan Celine Lin
PY  - 2024
UR  - https://arxiv.org/abs/2406.15765
ID  - 2406.15765
ER  -