Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2403.07825 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910655167070208 |
|---|---|
| author | Wang, Jianchen Gu, Zhouhong Zhu, Xiaoxuan Zhang, Lin Ye, Haoning Xiong, Zhuozhi Feng, Hongwei Xiao, Yanghua |
| author_facet | Wang, Jianchen Gu, Zhouhong Zhu, Xiaoxuan Zhang, Lin Ye, Haoning Xiong, Zhuozhi Feng, Hongwei Xiao, Yanghua |
| contents | Large Language Models have revolutionized numerous tasks with their remarkable efficacy. However, editing these models, crucial for rectifying outdated or erroneous information, often leads to a complex issue known as the ripple effect in the hidden space. While difficult to detect, this effect can significantly impede the efficacy of model editing tasks and deteriorate model performance. This paper addresses this scientific challenge by proposing a novel evaluation methodology, Graphical Impact Evaluation(GIE), which quantitatively evaluates the adaptations of the model and the subsequent impact of editing. Furthermore, we introduce the Selective Impact Revision(SIR), a model editing method designed to mitigate this ripple effect. Our comprehensive evaluations reveal that the ripple effect in the hidden space is a significant issue in all current model editing methods. However, our proposed methods, GIE and SIR, effectively identify and alleviate this issue, contributing to the advancement of LLM editing techniques. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_07825 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Efficiently Quantifying and Mitigating Ripple Effects in Model Editing Wang, Jianchen Gu, Zhouhong Zhu, Xiaoxuan Zhang, Lin Ye, Haoning Xiong, Zhuozhi Feng, Hongwei Xiao, Yanghua Computation and Language Large Language Models have revolutionized numerous tasks with their remarkable efficacy. However, editing these models, crucial for rectifying outdated or erroneous information, often leads to a complex issue known as the ripple effect in the hidden space. While difficult to detect, this effect can significantly impede the efficacy of model editing tasks and deteriorate model performance. This paper addresses this scientific challenge by proposing a novel evaluation methodology, Graphical Impact Evaluation(GIE), which quantitatively evaluates the adaptations of the model and the subsequent impact of editing. Furthermore, we introduce the Selective Impact Revision(SIR), a model editing method designed to mitigate this ripple effect. Our comprehensive evaluations reveal that the ripple effect in the hidden space is a significant issue in all current model editing methods. However, our proposed methods, GIE and SIR, effectively identify and alleviate this issue, contributing to the advancement of LLM editing techniques. |
| title | Efficiently Quantifying and Mitigating Ripple Effects in Model Editing |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2403.07825 |