LLMs Are Biased Towards Output Formats! Systematically Evaluating and Mitigating Output Format Bias of LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Long, Do Xuan, Ngoc, Hai Nguyen, Sim, Tiviatis, Dao, Hieu, Joty, Shafiq, Kawaguchi, Kenji, Chen, Nancy F., Kan, Min-Yen
Format:	Preprint
Published:	2024
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2408.08656