Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.18039 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908669703094272 |
|---|---|
| author | Bach, Thong Nguyen-Tang, Thanh Nguyen, Dung Le, Thao Minh Tran, Truyen |
| author_facet | Bach, Thong Nguyen-Tang, Thanh Nguyen, Dung Le, Thao Minh Tran, Truyen |
| contents | Fine-tuning Large Language Models (LLMs) for downstream tasks often compromises safety alignment, even when using parameter-efficient methods like LoRA. In this work, we uncover a notable property: fine-tuned models preserve the geometric structure of their loss landscapes concerning harmful content, regardless of the fine-tuning method employed. This suggests that safety behaviors are not erased but shifted to less influential regions of the parameter space. Building on this insight, we propose a curvature-aware alignment restoration method that leverages influence functions and second-order optimization to selectively increase loss on harmful inputs while preserving task performance. By navigating the shared geometry between base and fine-tuned models, our method discourages unsafe outputs while preserving task-relevant performance, avoiding full reversion and enabling precise, low-impact updates. Extensive evaluations across multiple model families and adversarial settings show that our approach efficiently reduces harmful responses while maintaining or even improving utility and few-shot learning performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_18039 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Curvature-Aware Safety Restoration In LLMs Fine-Tuning Bach, Thong Nguyen-Tang, Thanh Nguyen, Dung Le, Thao Minh Tran, Truyen Machine Learning Fine-tuning Large Language Models (LLMs) for downstream tasks often compromises safety alignment, even when using parameter-efficient methods like LoRA. In this work, we uncover a notable property: fine-tuned models preserve the geometric structure of their loss landscapes concerning harmful content, regardless of the fine-tuning method employed. This suggests that safety behaviors are not erased but shifted to less influential regions of the parameter space. Building on this insight, we propose a curvature-aware alignment restoration method that leverages influence functions and second-order optimization to selectively increase loss on harmful inputs while preserving task performance. By navigating the shared geometry between base and fine-tuned models, our method discourages unsafe outputs while preserving task-relevant performance, avoiding full reversion and enabling precise, low-impact updates. Extensive evaluations across multiple model families and adversarial settings show that our approach efficiently reduces harmful responses while maintaining or even improving utility and few-shot learning performance. |
| title | Curvature-Aware Safety Restoration In LLMs Fine-Tuning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2511.18039 |