Unveiling and Harnessing Hidden Attention Sinks: Enhancing Large Language Models without Training through Attention Calibration

Zhongzhi Yu, Zheng Wang, Yonggan Fu, Huihong Shi, Khalid Shaikh, Yingyan Celine Lin. Unveiling and Harnessing Hidden Attention Sinks: Enhancing Large Language Models without Training through Attention Calibration. In Forty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024. pages 57659-57677, OpenReview.net, 2024. [doi]

Authors

Zhongzhi Yu

This author has not been identified. Look up 'Zhongzhi Yu' in Google

Zheng Wang

This author has not been identified. Look up 'Zheng Wang' in Google

Yonggan Fu

This author has not been identified. Look up 'Yonggan Fu' in Google

Huihong Shi

This author has not been identified. Look up 'Huihong Shi' in Google

Khalid Shaikh

This author has not been identified. Look up 'Khalid Shaikh' in Google

Yingyan Celine Lin

This author has not been identified. Look up 'Yingyan Celine Lin' in Google