Quantifying Label-Induced Bias in Large Language Model Self- and Cross-Evaluations

This research investigates how label-induced bias distorts Large Language Model evaluations, revealing that perceived identity skews preference rankings. It ...

Level: advanced

By Unknown

Category: discussion