Promoting Equality in Large Language Models: Identifying and Mitigating the Implicit Bias based on Bayesian Theory

Yongxin Deng, Xihe Qiu, Xiaoyu Tan +6

Large language models (LLMs) are trained on extensive text corpora, which inevitably include biased information. Although techniques such as Affective Alignment can mitigate some n…