Related papers: KLAAD: Refining Attention Mechanisms to Reduce Societal Bias in Generative Language Models

KLAAD: Refining Attention Mechanisms to Reduce Societal Bias in Generative Language Models

URL: http://arxiv.org/abs/2507.19962v1
Date: Sat, 26 Jul 2025 14:24:19 GMT
Title: KLAAD: Refining Attention Mechanisms to Reduce Societal Bias in Generative Language Models
Authors: Seorin Kim, Dongyoung Lee, Jaejin Lee,
Abstract summary: Large language models (LLMs) often exhibit societal biases in their outputs, prompting ethical concerns regarding fairness and harm.<n>We propose KLAAD (KL-Attention Alignment Debiasing), an attention-based debiasing framework that implicitly aligns attention distributions between stereotypical and anti-stereotypical sentence pairs.<n> Experimental evaluation of KLAAD demonstrates improved bias mitigation on both the BBQ and BOLD benchmarks, with minimal impact on language modeling quality.
Score: 1.649505438157608
License: http://creativecommons.org/licenses/by-nc-sa/4.0/
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.

Related papers

Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling [49.41422138354821]
We propose a principled reward modeling framework that integrates non-negative factor analysis into the Bradley-Terry preference model.<n>BNRM represents rewards through a sparse, non-negative latent factor generative process.<n>We show that BNRM substantially mitigates reward over-optimization, improves robustness under distribution shifts, and yields more interpretable reward decompositions than strong baselines.
arXiv Detail & Related papers (2026-02-11T08:14:11Z)
Beacon: Single-Turn Diagnosis and Mitigation of Latent Sycophancy in Large Language Models [0.0]
Large language models internalize a structural trade-off between truthfulness and obsequious flattery.<n>This latent bias, known as sycophancy, manifests as a preference for user agreement over principled reasoning.<n>We introduce Beacon, a single-turn forced-choice benchmark that isolates this bias independent of conversational context.
arXiv Detail & Related papers (2025-10-19T06:36:57Z)
Mitigating Spurious Correlations in LLMs via Causality-Aware Post-Training [57.03005244917803]
Large language models (LLMs) often fail on out-of-distribution (OOD) samples due to spurious correlations acquired during pre-training.<n>Here, we aim to mitigate such spurious correlations through causality-aware post-training (CAPT)<n> Experiments on the formal causal inference benchmark CLadder and the logical reasoning dataset PrOntoQA show that 3B-scale language models fine-tuned with CAPT can outperform both traditional SFT and larger LLMs on in-distribution (ID) and OOD tasks.
arXiv Detail & Related papers (2025-06-11T06:30:28Z)
Class-Conditional Distribution Balancing for Group Robust Classification [11.525201208566925]
Spurious correlations that lead models to correct predictions for the wrong reasons pose a critical challenge for robust real-world generalization.<n>We offer a novel perspective by reframing the spurious correlations as imbalances or mismatches in class-conditional distributions.<n>We propose a simple yet effective robust learning method that eliminates the need for both bias annotations and predictions.
arXiv Detail & Related papers (2025-04-24T07:15:53Z)
A Causal Adjustment Module for Debiasing Scene Graph Generation [28.44150555570101]
We employ causal inference techniques to model the causality among skewed distributions.<n>Our method enables the composition of zero-shot relationships, thereby enhancing the model's ability to recognize such relationships.
arXiv Detail & Related papers (2025-03-22T20:44:01Z)
Rethinking Relation Extraction: Beyond Shortcuts to Generalization with a Debiased Benchmark [53.876493664396506]
Benchmarks are crucial for evaluating machine learning algorithm performance, facilitating comparison and identifying superior solutions.<n>This paper addresses the issue of entity bias in relation extraction tasks, where models tend to rely on entity mentions rather than context.<n>We propose a debiased relation extraction benchmark DREB that breaks the pseudo-correlation between entity mentions and relation types through entity replacement.<n>To establish a new baseline on DREB, we introduce MixDebias, a debiasing method combining data-level and model training-level techniques.
arXiv Detail & Related papers (2025-01-02T17:01:06Z)
Bias Begets Bias: The Impact of Biased Embeddings on Diffusion Models [0.0]
Text-to-Image (TTI) systems have come under increased scrutiny for social biases. We investigate embedding spaces as a source of bias for TTI models. We find that biased multimodal embeddings like CLIP can result in lower alignment scores for representationally balanced TTI models.
arXiv Detail & Related papers (2024-09-15T01:09:55Z)
Identifying and Mitigating Social Bias Knowledge in Language Models [52.52955281662332]
We propose a novel debiasing approach, Fairness Stamp (FAST), which enables fine-grained calibration of individual social biases.<n>FAST surpasses state-of-the-art baselines with superior debiasing performance.<n>This highlights the potential of fine-grained debiasing strategies to achieve fairness in large language models.
arXiv Detail & Related papers (2024-08-07T17:14:58Z)
Debiasing Vision-Language Models via Biased Prompts [79.04467131711775]
We propose a general approach for debiasing vision-language foundation models by projecting out biased directions in the text embedding. We show that debiasing only the text embedding with a calibrated projection matrix suffices to yield robust classifiers and fair generative models.
arXiv Detail & Related papers (2023-01-31T20:09:33Z)
How robust are pre-trained models to distribution shift? [82.08946007821184]
We show how spurious correlations affect the performance of popular self-supervised learning (SSL) and auto-encoder based models (AE) We develop a novel evaluation scheme with the linear head trained on out-of-distribution (OOD) data, to isolate the performance of the pre-trained models from a potential bias of the linear head used for evaluation.
arXiv Detail & Related papers (2022-06-17T16:18:28Z)
General Greedy De-bias Learning [163.65789778416172]
We propose a General Greedy De-bias learning framework (GGD), which greedily trains the biased models and the base model like gradient descent in functional space. GGD can learn a more robust base model under the settings of both task-specific biased models with prior knowledge and self-ensemble biased model without prior knowledge.
arXiv Detail & Related papers (2021-12-20T14:47:32Z)
Fairness by Explicability and Adversarial SHAP Learning [0.0]
We propose a new definition of fairness that emphasises the role of an external auditor and model explicability. We develop a framework for mitigating model bias using regularizations constructed from the SHAP values of an adversarial surrogate model. We demonstrate our approaches using gradient and adaptive boosting on: a synthetic dataset, the UCI Adult (Census) dataset and a real-world credit scoring dataset.
arXiv Detail & Related papers (2020-03-11T14:36:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.