Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2502.17601 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910843167309824 |
|---|---|
| author | Bartoszcze, Lukasz Munshi, Sarthak Sukidi, Bryan Yen, Jennifer Yang, Zejia Williams-King, David Le, Linh Asuzu, Kosi Maple, Carsten |
| author_facet | Bartoszcze, Lukasz Munshi, Sarthak Sukidi, Bryan Yen, Jennifer Yang, Zejia Williams-King, David Le, Linh Asuzu, Kosi Maple, Carsten |
| contents | Large-language models are capable of completing a variety of tasks, but remain unpredictable and intractable. Representation engineering seeks to resolve this problem through a new approach utilizing samples of contrasting inputs to detect and edit high-level representations of concepts such as honesty, harmfulness or power-seeking. We formalize the goals and methods of representation engineering to present a cohesive picture of work in this emerging discipline. We compare it with alternative approaches, such as mechanistic interpretability, prompt-engineering and fine-tuning. We outline risks such as performance decrease, compute time increases and steerability issues. We present a clear agenda for future research to build predictable, dynamic, safe and personalizable LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_17601 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Representation Engineering for Large-Language Models: Survey and Research Challenges Bartoszcze, Lukasz Munshi, Sarthak Sukidi, Bryan Yen, Jennifer Yang, Zejia Williams-King, David Le, Linh Asuzu, Kosi Maple, Carsten Artificial Intelligence Large-language models are capable of completing a variety of tasks, but remain unpredictable and intractable. Representation engineering seeks to resolve this problem through a new approach utilizing samples of contrasting inputs to detect and edit high-level representations of concepts such as honesty, harmfulness or power-seeking. We formalize the goals and methods of representation engineering to present a cohesive picture of work in this emerging discipline. We compare it with alternative approaches, such as mechanistic interpretability, prompt-engineering and fine-tuning. We outline risks such as performance decrease, compute time increases and steerability issues. We present a clear agenda for future research to build predictable, dynamic, safe and personalizable LLMs. |
| title | Representation Engineering for Large-Language Models: Survey and Research Challenges |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2502.17601 |