Saved in:
Bibliographic Details
Main Authors: Shen, Hanwen, Ying, Ting, Lu, Jiajie, Wang, Shanshan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.13683
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911600099721216
author Shen, Hanwen
Ying, Ting
Lu, Jiajie
Wang, Shanshan
author_facet Shen, Hanwen
Ying, Ting
Lu, Jiajie
Wang, Shanshan
contents Although debiased large language models (LLMs) excel at handling known or low-bias prompts, they often fail on unfamiliar and high-bias prompts. We demonstrate via out-of-distribution (OOD) detection that these high-bias prompts cause a distribution shift, degrading static model performance. To enable real-time correction, we propose CAP-TTA, a test-time adaptation framework. CAP-TTA triggers context-aware LoRA updates only when a bias-risk score exceeds a set threshold. By utilizing an offline precomputed diagonal preconditioner, it ensures fast and stable optimization. Across multiple benchmarks and human evaluations, CAP-TTA effectively reduces toxicity/bias score with significantly lower latency than standard optimization methods (e.g., AdamW or SGD). Furthermore, it prevents catastrophic forgetting, and substantially improves narrative fluency over state-of-the-art baselines without compromising debiasing performance.
format Preprint
id arxiv_https___arxiv_org_abs_2603_13683
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation
Shen, Hanwen
Ying, Ting
Lu, Jiajie
Wang, Shanshan
Computation and Language
Artificial Intelligence
Computers and Society
Although debiased large language models (LLMs) excel at handling known or low-bias prompts, they often fail on unfamiliar and high-bias prompts. We demonstrate via out-of-distribution (OOD) detection that these high-bias prompts cause a distribution shift, degrading static model performance. To enable real-time correction, we propose CAP-TTA, a test-time adaptation framework. CAP-TTA triggers context-aware LoRA updates only when a bias-risk score exceeds a set threshold. By utilizing an offline precomputed diagonal preconditioner, it ensures fast and stable optimization. Across multiple benchmarks and human evaluations, CAP-TTA effectively reduces toxicity/bias score with significantly lower latency than standard optimization methods (e.g., AdamW or SGD). Furthermore, it prevents catastrophic forgetting, and substantially improves narrative fluency over state-of-the-art baselines without compromising debiasing performance.
title Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation
topic Computation and Language
Artificial Intelligence
Computers and Society
url https://arxiv.org/abs/2603.13683