Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation
Hanwen Shen Ting Ying Jiajie Lu Shanshan Wang · 2026
Read Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation by Hanwen Shen Ting Ying Jiajie Lu Shanshan Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-13683.txt.
Description
Hanwen Shen and co-authors propose CAP-TTA for out-of-distribution debiasing in narrative generation. Debiased LLMs handle known or low-bias prompts well but often fail on unfamiliar, high-bias ones. The team shows those prompts cause a distribution shift that degrades static model performance. CAP-TTA triggers context-aware LoRA updates only when a bias-risk score passes a set threshold. An offline precomputed diagonal preconditioner keeps optimization fast and stable. Across benchmarks and human evaluations it cut toxicity and bias with lower latency than standard methods.
Catalog record
| CREATOR | Hanwen Shen Ting Ying Jiajie Lu Shanshan Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69191 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.13683