Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.14004 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917405664477184 |
|---|---|
| author | Zhang, Hengyuan Zhang, Zhihao Wang, Mingyang Su, Zunhai Wang, Yiwei Wang, Qianli Yuan, Shuzhou Nie, Ercong Duan, Xufeng Han, Feijiang Xue, Qibo Yu, Zeping Shang, Chenming Liang, Xiao Xiong, Jing Shen, Hui Tao, Chaofan Liu, Zhengwu Jin, Senjie Xi, Zhiheng Zhang, Dongdong Ananiadou, Sophia Gui, Tao Xie, Ruobing So, Hayden Kwok-Hay Schütze, Hinrich Huang, Xuanjing Zhang, Qi Wong, Ngai |
| author_facet | Zhang, Hengyuan Zhang, Zhihao Wang, Mingyang Su, Zunhai Wang, Yiwei Wang, Qianli Yuan, Shuzhou Nie, Ercong Duan, Xufeng Han, Feijiang Xue, Qibo Yu, Zeping Shang, Chenming Liang, Xiao Xiong, Jing Shen, Hui Tao, Chaofan Liu, Zhengwu Jin, Senjie Xi, Zhiheng Zhang, Dongdong Ananiadou, Sophia Gui, Tao Xie, Ruobing So, Hayden Kwok-Hay Schütze, Hinrich Huang, Xuanjing Zhang, Qi Wong, Ngai |
| contents | Mechanistic Interpretability (MI) has emerged as a vital approach to demystify the opaque decision-making of Large Language Models (LLMs). However, existing reviews primarily treat MI as an observational science, summarizing analytical insights while lacking a systematic framework for actionable intervention. To bridge this gap, we present a practical survey structured around the pipeline: "Locate, Steer, and Improve." We formally categorize Localizing (diagnosis) and Steering (intervention) methods based on specific Interpretable Objects to establish a rigorous intervention protocol. Furthermore, we demonstrate how this framework enables tangible improvements in Alignment, Capability, and Efficiency, effectively operationalizing MI as an actionable methodology for model optimization. The curated paper list of this work is available at https://github.com/rattlesnakey/Awesome-Actionable-MI-Survey. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_14004 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models Zhang, Hengyuan Zhang, Zhihao Wang, Mingyang Su, Zunhai Wang, Yiwei Wang, Qianli Yuan, Shuzhou Nie, Ercong Duan, Xufeng Han, Feijiang Xue, Qibo Yu, Zeping Shang, Chenming Liang, Xiao Xiong, Jing Shen, Hui Tao, Chaofan Liu, Zhengwu Jin, Senjie Xi, Zhiheng Zhang, Dongdong Ananiadou, Sophia Gui, Tao Xie, Ruobing So, Hayden Kwok-Hay Schütze, Hinrich Huang, Xuanjing Zhang, Qi Wong, Ngai Computation and Language Mechanistic Interpretability (MI) has emerged as a vital approach to demystify the opaque decision-making of Large Language Models (LLMs). However, existing reviews primarily treat MI as an observational science, summarizing analytical insights while lacking a systematic framework for actionable intervention. To bridge this gap, we present a practical survey structured around the pipeline: "Locate, Steer, and Improve." We formally categorize Localizing (diagnosis) and Steering (intervention) methods based on specific Interpretable Objects to establish a rigorous intervention protocol. Furthermore, we demonstrate how this framework enables tangible improvements in Alignment, Capability, and Efficiency, effectively operationalizing MI as an actionable methodology for model optimization. The curated paper list of this work is available at https://github.com/rattlesnakey/Awesome-Actionable-MI-Survey. |
| title | Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.14004 |