KLAAD: Refining Attention Mechanisms to Reduce Societal Bias in Generative Language Models
Year: 2025
Type: preprint
Abstract: Large language models (LLMs) often exhibit societal biases in their outputs, prompting ethical concerns regarding fairness and harm. In this work, we propose KLAAD (KL-Attention Alignment Debiasing), an attention-based debiasing framework that implicitly aligns attention distributions between stereotypical and anti-stereotypical sentence pairs without directly modifying model weights. KLAAD introduces a composite training objective combining Cross-Entropy, KL divergence, and Triplet losses, guiding the model to consistently attend across biased and unbiased contexts while preserving fluency and coherence. Experimental evaluation of KLAAD demonstrates improved bias mitigation on both the BBQ and BOLD benchmarks, with minimal impact on language modeling quality. The results indicate that attention-level alignment offers a principled solution for mitigating bias in generative language models.
Source: arXiv (Cornell University)
Authors: Kim, Seorin, Lee, Dongyoung, Lee, Jaejin
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
SDG: Reduced inequalities
Open Access status: green
OpenAlex ID: W6929326803
Cites:
Cited by:
Related to: