ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Training

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Han, Feijiang, Yu, Xiaodong, Tang, Jianheng, Rao, Delip, Du, Weihua, Ungar, Lyle
Format:	Preprint
Published:	2025
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2505.11739