---
title: "Programming Refusal with Conditional Activation Steering (2024)"
description: "Lee et al. propose Conditional Activation Steering (CAST), which makes steering depend on the input. 91 references, cited by 41 papers here."
authors: ["Bruce W. Lee", "Inkit Padhi", "Karthikeyan Natesan Ramamurthy", "Erik Miehling", "Pierre Dognin", "Manish Nagireddy", "Amit Dhurandhar"]
year: 2024
venue: "arXiv:2409.05907"
source: "https://arxiv.org/abs/2409.05907"
doi: "10.48550/arxiv.2409.05907"
arxiv: "2409.05907"
licence: "CC BY-SA 4.0 (arXiv)"
licence_url: "http://creativecommons.org/licenses/by-sa/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/lee-2024-programming-refusal-conditional-activation-steering/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lee-2024-programming-refusal-conditional-activation-steering/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2024](https://md.research.elamur.ai/free/en/papers/2024/page/9/) / Programming Refusal with Conditional Activation Steering

# Programming Refusal with Conditional Activation Steering

**Authors:** [Bruce W. Lee](https://md.research.elamur.ai/free/en/authors/bruce-w-lee/), [Inkit Padhi](https://md.research.elamur.ai/free/en/authors/inkit-padhi/), [Karthikeyan Natesan Ramamurthy](https://md.research.elamur.ai/free/en/authors/karthikeyan-natesan-ramamurthy/), [Erik Miehling](https://md.research.elamur.ai/free/en/authors/erik-miehling/), [Pierre Dognin](https://md.research.elamur.ai/free/en/authors/pierre-dognin/), [Manish Nagireddy](https://md.research.elamur.ai/free/en/authors/manish-nagireddy/) and Amit Dhurandhar · **Year:** 2024 · **Venue:** arXiv:2409.05907 · **Licence:** [CC BY-SA 4.0 (arXiv)](http://creativecommons.org/licenses/by-sa/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2409.05907) · [PDF](https://arxiv.org/pdf/2409.05907) · [DOI 10.48550/arxiv.2409.05907](https://doi.org/10.48550/arxiv.2409.05907) · [arXiv 2409.05907](https://arxiv.org/abs/2409.05907) · [OpenAlex](https://openalex.org/W4403590208)

**Citations:** 1 citations (OpenAlex, 2026-10-14) · [41](https://md.research.elamur.ai/free/en/papers/lee-2024-programming-refusal-conditional-activation-steering/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Activation steering adds a vector to an LLM's hidden states during inference to change its behavior. Existing methods apply that vector to every input, so a refusal vector makes the model decline harmless requests as often as harmful ones. Lee et al. propose Conditional Activation Steering (CAST), which makes steering depend on the input. In addition to the usual behavior vector (for example, one that induces refusal), they extract a condition vector for each layer. A condition vector captures the activation pattern of a prompt category, such as adult content or discrimination. At inference, the hidden state is projected onto the condition vector. If the cosine similarity between the state and its projection passes a threshold, the behavior vector is added for the rest of the generation. Conditions can be combined logically, for instance refusing only hate speech and adult content, or refusing anything unrelated to legal advice. Experiments with sorry-bench and Alpaca prompts on several open chat models show that CAST refuses the targeted prompts while keeping harmless refusals low. It needs no weight optimization and stays as cheap as plain steering, which makes it practical for content moderation and domain-specific assistants.

## Key points
- **Core mechanism:** the behavior vector is scaled by α and gated by a step function. The step function fires when the cosine score comparing the hidden state with its condition-vector projection exceeds θ.
- **Calibration:** the condition layer, threshold and comparison direction are chosen by grid search that maximizes F1 between positive and negative examples. The search covers only the first half of the model's layers.
- **Data and models:** data come from sorry-bench (sorry-bench-202406) and a cleaned Alpaca set. Models include:
  - Qwen1.5-1.8B-Chat and Qwen1.5-32B-Chat
  - Llama-2-13b-chat and Llama-3.1-8B-Instruct
  - NeuralDaredevil-8B-abliterated and Hermes-2-Pro-Llama-3-8B
  - OLMo-7B-SFT, zephyr-7b-beta and h2o-danube3-4b-chat
- **Refusal measurement:** the classifier protectai/distilroberta-base-rejection-v1 is combined with a case-insensitive keyword list. A response counts as a refusal if either check flags it.
- **Release and setup:** the open-source toolkit is published at github.com/IBM/activation-steering. Experiments ran on an NVIDIA A100-SXM4-80GB with PyTorch 2.3.0 and Transformers 4.43.3.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Pennsylvania](https://md.research.elamur.ai/free/en/universities/university-of-pennsylvania/) — University, Philadelphia, United States · 241 works · 66,904 citations
- [IBM](https://md.research.elamur.ai/free/en/companies/ibm/) — Company, Armonk, United States · 163 works · 73,139 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 19,503 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 18,805 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,959 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,789 works

## Cite

Cite as: Bruce W. Lee, Inkit Padhi, Karthikeyan Natesan Ramamurthy, Erik Miehling, Pierre Dognin, Manish Nagireddy and Amit Dhurandhar. 2024. Programming Refusal with Conditional Activation Steering. arXiv:2409.05907.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{lee2024programming,
  title = {{Programming Refusal with Conditional Activation Steering}},
  author = {Bruce W. Lee, Inkit Padhi, Karthikeyan Natesan Ramamurthy, Erik Miehling, Pierre Dognin, Manish Nagireddy and Amit Dhurandhar},
  year = {2024},
  doi = {10.48550/arxiv.2409.05907},
  eprint = {2409.05907},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2409.05907}
}
```

- [← previous On the Self-Verification Limitations of Large Language Models on Reasoning and Planning Tasks](https://md.research.elamur.ai/free/en/papers/stechly-2024-self-verification-limitations-large-language/)
- [next → Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models](https://md.research.elamur.ai/free/en/papers/ormazabal-2024-reka-core-flash-edge-series/)
- [↑ back to papers of 2024](https://md.research.elamur.ai/free/en/papers/2024/page/9/)

## References

91 references, 35 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Dyah Adila, Shuai Zhang, Boran Han, and Yuyang Wang. Discovering bias in latent space: An unsupervised debiasing approach. arXiv preprint arXiv:2406.03631, 2024. — [arXiv:2406.03631](https://arxiv.org/abs/2406.03631) [find](https://scholar.google.com/scholar?q=Dyah+Adila%2C+Shuai+Zhang%2C+Boran+Han%2C+and+Yuyang+Wang%2E+Discovering+bias+in+latent+space%3A+An+unsupervised+debiasing+approach%2E+arXiv+preprint+arXiv%3A2406%2E03631%2C+2024%2E+%E2%80%94+%5BarXiv%3A2406%2E03631%5D%28https%3A%2F%2Farxiv%2Eorg)
2. Alan Ansell, Ivan Vulić, Hannah Sterz, Anna Korhonen, and Edoardo M Ponti. Scaling sparse fine-tuning to large language models. arXiv preprint arXiv:2401.16405, 2024. — [arXiv:2401.16405](https://arxiv.org/abs/2401.16405) [find](https://scholar.google.com/scholar?q=Alan+Ansell%2C+Ivan+Vulic%CC%81%2C+Hannah+Sterz%2C+Anna+Korhonen%2C+and+Edoardo+M+Ponti%2E+Scaling+sparse+fine%2Dtuning+to+large+language+models%2E+arXiv+preprint+arXiv%3A2401%2E16405%2C+2024%2E+%E2%80%94+%5BarXiv%3A2401%2E16405%5D%28https%3A%2F%2Farx)
3. Usman Anwar, Abulhair Saparov, Javier Rando, Daniel Paleka, Miles Turpin, Peter Hase, Ekdeep Singh Lubana, Erik Jenner, Stephen Casper, Oliver Sourbut, et al. [Foundational challenges in assuring alignment and safety of large language models](https://md.research.elamur.ai/free/en/papers/anwar-2024-foundational-challenges-assuring-alignment-safety/). arXiv preprint arXiv:2404.09932, 2024.
4. Andy Arditi, Oscar Obeso, Aaquib Syed, Daniel Paleka, Nina Rimsky, Wes Gurnee, and Neel Nanda. [Refusal in language models is mediated by a single direction](https://md.research.elamur.ai/free/en/papers/arditi-2024-refusal-language-models-mediated-single/), 2024.
5. Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, et al. A general language assistant as a laboratory for alignment. arXiv preprint arXiv:2112.00861, 2021. [find](https://scholar.google.com/scholar?q=Amanda+Askell%2C+Yuntao+Bai%2C+Anna+Chen%2C+Dawn+Drain%2C+Deep+Ganguli%2C+Tom+Henighan%2C+Andy+Jones%2C+Nicholas+Joseph%2C+Ben+Mann%2C+Nova+DasSarma%2C+et+al%2E+A+general+language+assistant+as+a+laboratory+for+alignment%2E+a)
6. Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, P… — [Qwen Technical Report](https://md.research.elamur.ai/free/en/papers/bai-2023-qwen-technical-report/)
7. Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. [Constitutional ai: Harmlessness from ai feedback](https://md.research.elamur.ai/free/en/papers/bai-2022-constitutional-ai-harmlessness-ai-feedback/). arXiv preprint arXiv:2212.08073, 2022.
8. Sarah Ball, Frauke Kreuter, and Nina Rimsky. [Understanding jailbreak success: A study of latent space dynamics in large language models](https://md.research.elamur.ai/free/en/papers/ball-2026-understanding-jailbreak-success-study-latent/). arXiv preprint arXiv:2406.09289, 2024.
9. Faeze Brahman, Sachin Kumar, Vidhisha Balachandran, Pradeep Dasigi, Valentina Pyatkin, Abhilasha Ravichander, Sarah Wiegreffe, Nouha Dziri, Khyathi Chandu, Jack Hessel, et al. [The art of saying no: Contextual noncompliance in language models](https://md.research.elamur.ai/free/en/papers/brahman-2024-art-saying-no-contextual-noncompliance/). arXiv preprint arXiv:2407.12043, 2024.
10. Yuanpu Cao, Tianrong Zhang, Bochuan Cao, Ziyi Yin, Lu Lin, Fenglong Ma, and Jinghui Chen. Personalized steering of large language models: Versatile steering vectors through bi-directional preference optimization. arXiv preprint arXiv:2406.00045, 2024. — [arXiv:2406.00045](https://arxiv.org/abs/2406.00045) [find](https://scholar.google.com/scholar?q=Yuanpu+Cao%2C+Tianrong+Zhang%2C+Bochuan+Cao%2C+Ziyi+Yin%2C+Lu+Lin%2C+Fenglong+Ma%2C+and+Jinghui+Chen%2E+Personalized+steering+of+large+language+models%3A+Versatile+steering+vectors+through+bi%2Ddirectional+preference+o)
11. Ziwei Chai, Guoyin Wang, Jing Su, Tianjie Zhang, Xuanwen Huang, Xuwu Wang, Jingjing Xu, Jianbo Yuan, Hongxia Yang, Fei Wu, et al. An expert is worth one token: Synergizing multiple expert llms as generalist via expert token routing. arXiv preprint arXiv:2403.16854, 2024. [find](https://scholar.google.com/scholar?q=Ziwei+Chai%2C+Guoyin+Wang%2C+Jing+Su%2C+Tianjie+Zhang%2C+Xuanwen+Huang%2C+Xuwu+Wang%2C+Jingjing+Xu%2C+Jianbo+Yuan%2C+Hongxia+Yang%2C+Fei+Wu%2C+et+al%2E+An+expert+is+worth+one+token%3A+Synergizing+multiple+expert+llms+as+gene)
12. Inyoung Cheong, King Xia, KJ Kevin Feng, Quan Ze Chen, and Amy X Zhang. (a) i am not a lawyer, but...: Engaging legal experts towards responsible llm policies for legal advice. In The 2024 ACM Conference on Fairness, Accountability, and Transparency, pp. 2454–2469, 2024. [find](https://scholar.google.com/scholar?q=Inyoung+Cheong%2C+King+Xia%2C+KJ+Kevin+Feng%2C+Quan+Ze+Chen%2C+and+Amy+X+Zhang%2E+%28a%29+i+am+not+a+lawyer%2C+but%2E%2E%2E%3A+Engaging+legal+experts+towards+responsible+llm+policies+for+legal+advice%2E+In+The+2024+ACM+Confere)
13. Justin Cui, Wei-Lin Chiang, Ion Stoica, and Cho-Jui Hsieh. [Or-bench: An over-refusal benchmark for large language models](https://md.research.elamur.ai/free/en/papers/cui-2024-bench-over-refusal-benchmark-large/). arXiv preprint arXiv:2405.20947, 2024.
14. Nirjhar Das, Souradip Chakraborty, Aldo Pacchiano, and Sayak Ray Chowdhury. Provably sample efficient rlhf via active preference optimization. arXiv preprint arXiv:2402.10500, 2024. — [arXiv:2402.10500](https://arxiv.org/abs/2402.10500) [find](https://scholar.google.com/scholar?q=Nirjhar+Das%2C+Souradip+Chakraborty%2C+Aldo+Pacchiano%2C+and+Sayak+Ray+Chowdhury%2E+Provably+sample+efficient+rlhf+via+active+preference+optimization%2E+arXiv+preprint+arXiv%3A2402%2E10500%2C+2024%2E+%E2%80%94+%5BarXiv%3A2402%2E1050)
15. Jasper Dekoninck, Marc Fischer, Luca Beurer-Kellner, and Martin Vechev. Controlled text generation via language model arithmetic. arXiv preprint arXiv:2311.14479, 2023. [find](https://scholar.google.com/scholar?q=Jasper+Dekoninck%2C+Marc+Fischer%2C+Luca+Beurer%2DKellner%2C+and+Martin+Vechev%2E+Controlled+text+generation+via+language+model+arithmetic%2E+arXiv+preprint+arXiv%3A2311%2E14479%2C+2023%2E)
16. Nelson Elhage, Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, et al. A mathematical framework for transformer circuits. Transformer Circuits Thread, 1(1):12, 2021. [find](https://scholar.google.com/scholar?q=Nelson+Elhage%2C+Neel+Nanda%2C+Catherine+Olsson%2C+Tom+Henighan%2C+Nicholas+Joseph%2C+Ben+Mann%2C+Amanda+Askell%2C+Yuntao+Bai%2C+Anna+Chen%2C+Tom+Conerly%2C+et+al%2E+A+mathematical+framework+for+transformer+circuits%2E+Trans)
17. Shangbin Feng, Taylor Sorensen, Yuhan Liu, Jillian Fisher, Chan Young Park, Yejin Choi, and Yulia Tsvetkov. [Modular pluralism: Pluralistic alignment via multi-llm collaboration](https://md.research.elamur.ai/free/en/papers/feng-2024-modular-pluralism-pluralistic-alignment-multi/). arXiv preprint arXiv:2406.15951, 2024.
18. Asma Ghandeharioun, Ann Yuan, Marius Guerard, Emily Reif, Michael A Lepori, and Lucas Dixon. Who’s asking? user personas and the mechanics of latent misalignment. arXiv preprint arXiv:2406.12094, 2024. — [arXiv:2406.12094](https://arxiv.org/abs/2406.12094) [find](https://scholar.google.com/scholar?q=Asma+Ghandeharioun%2C+Ann+Yuan%2C+Marius+Guerard%2C+Emily+Reif%2C+Michael+A+Lepori%2C+and+Lucas+Dixon%2E+Who%E2%80%99s+asking%3F+user+personas+and+the+mechanics+of+latent+misalignment%2E+arXiv+preprint+arXiv%3A2406%2E12094%2C+2024)
19. Dirk Groeneveld, Iz Beltagy, Pete Walsh, Akshita Bhagia, Rodney Kinney, Oyvind Tafjord, Ananya Harsh Jha, Hamish Ivison, Ian Magnusson, Yizhong Wang, Shane Arora, David Atkinson, Russell Authur, Khyathi Chandu, Arman Cohan, Jennifer Dumas, Yanai Elazar, Yuling Gu, Jack Hessel, Tushar Khot, William M… [find](https://scholar.google.com/scholar?q=Dirk+Groeneveld%2C+Iz+Beltagy%2C+Pete+Walsh%2C+Akshita+Bhagia%2C+Rodney+Kinney%2C+Oyvind+Tafjord%2C+Ananya+Harsh+Jha%2C+Hamish+Ivison%2C+Ian+Magnusson%2C+Yizhong+Wang%2C+Shane+Arora%2C+David+Atkinson%2C+Russell+Authur%2C+Khyat)
20. Wes Gurnee and Max Tegmark. Language models represent space and time. arXiv preprint arXiv:2310.02207, 2023. [find](https://scholar.google.com/scholar?q=Wes+Gurnee+and+Max+Tegmark%2E+Language+models+represent+space+and+time%2E+arXiv+preprint+arXiv%3A2310%2E02207%2C+2023%2E)
21. Chi Han, Jialiang Xu, Manling Li, Yi Fung, Chenkai Sun, Nan Jiang, Tarek Abdelzaher, and Heng Ji. [Word embeddings are steers for language models](https://md.research.elamur.ai/free/en/papers/han-2024-word-embeddings-are-steers-language/), 2024. URL https://arxiv.org/abs/ 2305.12798. [↗](https://arxiv.org/abs/)
22. Benjamin David Hayum, Quentin Feuillade Montixi, and Yixuan Li. How does rlhf shift behavior distributions? distinguishability and steerability. Jerry Zhi-Yang He, Sashrika Pandey, Mariah L Schrum, and Anca Dragan. Cos: Enhancing personalization and mitigating bias with context steering. arXiv prepr… — [arXiv:2405.01768](https://arxiv.org/abs/2405.01768) [find](https://scholar.google.com/scholar?q=Benjamin+David+Hayum%2C+Quentin+Feuillade+Montixi%2C+and+Yixuan+Li%2E+How+does+rlhf+shift+behavior+distributions%3F+distinguishability+and+steerability%2E+Jerry+Zhi%2DYang+He%2C+Sashrika+Pandey%2C+Mariah+L+Schrum%2C+an)
23. Zihao He, Siyi Guo, Ashwin Rao, and Kristina Lerman. Whose emotions and moral sentiments do language models reflect? arXiv preprint arXiv:2402.11114, 2024b. [find](https://scholar.google.com/scholar?q=Zihao+He%2C+Siyi+Guo%2C+Ashwin+Rao%2C+and+Kristina+Lerman%2E+Whose+emotions+and+moral+sentiments+do+language+models+reflect%3F+arXiv+preprint+arXiv%3A2402%2E11114%2C+2024b%2E)
24. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. [Measuring mathematical problem solving with the math dataset](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-measuring-mathematical-problem-solving-math/). NeurIPS, 2021.
25. Chia-Yi Hsu, Yu-Lin Tsai, Chih-Hsun Lin, Pin-Yu Chen, Chia-Mu Yu, and Chun-Ying Huang. Safe lora: the silver lining of reducing safety risks when fine-tuning large language models. arXiv preprint arXiv:2405.16833, 2024. — [arXiv:2405.16833](https://arxiv.org/abs/2405.16833) [find](https://scholar.google.com/scholar?q=Chia%2DYi+Hsu%2C+Yu%2DLin+Tsai%2C+Chih%2DHsun+Lin%2C+Pin%2DYu+Chen%2C+Chia%2DMu+Yu%2C+and+Chun%2DYing+Huang%2E+Safe+lora%3A+the+silver+lining+of+reducing+safety+risks+when+fine%2Dtuning+large+language+models%2E+arXiv+preprint+arXi)
26. Zhanhao Hu, Julien Piet, Geng Zhao, Jiantao Jiao, and David Wagner. Toxicity detection for free, 2024. URL https://arxiv.org/abs/2405.18822. — [arXiv:2405.18822](https://arxiv.org/abs/2405.18822) [↗](https://arxiv.org/abs/2405.18822)
27. Joel Jang, Seonghyeon Ye, and Minjoon Seo. Can large language models truly understand prompts? a case study with negated prompts. In Transfer learning for natural language processing workshop, pp. 52–62. PMLR, 2023. [find](https://scholar.google.com/scholar?q=Joel+Jang%2C+Seonghyeon+Ye%2C+and+Minjoon+Seo%2E+Can+large+language+models+truly+understand+prompts%3F+a+case+study+with+negated+prompts%2E+In+Transfer+learning+for+natural+language+processing+workshop%2C+pp%2E+52%E2%80%93)
28. Ole Jorgensen, Dylan Cope, Nandi Schoots, and Murray Shanahan. Improving activation steering in language models with mean-centring. arXiv preprint arXiv:2312.03813, 2023. — [arXiv:2312.03813](https://arxiv.org/abs/2312.03813) [find](https://scholar.google.com/scholar?q=Ole+Jorgensen%2C+Dylan+Cope%2C+Nandi+Schoots%2C+and+Murray+Shanahan%2E+Improving+activation+steering+in+language+models+with+mean%2Dcentring%2E+arXiv+preprint+arXiv%3A2312%2E03813%2C+2023%2E+%E2%80%94+%5BarXiv%3A2312%2E03813%5D%28https%3A%2F%2F)
29. Kai Konen, Sophie Jentzsch, Diaoulé Diallo, Peer Schütt, Oliver Bensch, Roxanne El Baff, Dominik Opitz, and Tobias Hecking. Style vectors for steering generative large language model. arXiv preprint arXiv:2402.01618, 2024. — [arXiv:2402.01618](https://arxiv.org/abs/2402.01618) [find](https://scholar.google.com/scholar?q=Kai+Konen%2C+Sophie+Jentzsch%2C+Diaoul%C3%A9+Diallo%2C+Peer+Sch%C3%BCtt%2C+Oliver+Bensch%2C+Roxanne+El+Baff%2C+Dominik+Opitz%2C+and+Tobias+Hecking%2E+Style+vectors+for+steering+generative+large+language+model%2E+arXiv+preprint+a)
30. Lingkai Kong, Haorui Wang, Wenhao Mu, Yuanqi Du, Yuchen Zhuang, Yifei Zhou, Yue Song, Rongzhi Zhang, Kai Wang, and Chao Zhang. Aligning large language models with representation editing: A control perspective. arXiv preprint arXiv:2406.05954, 2024. — [arXiv:2406.05954](https://arxiv.org/abs/2406.05954) [find](https://scholar.google.com/scholar?q=Lingkai+Kong%2C+Haorui+Wang%2C+Wenhao+Mu%2C+Yuanqi+Du%2C+Yuchen+Zhuang%2C+Yifei+Zhou%2C+Yue+Song%2C+Rongzhi+Zhang%2C+Kai+Wang%2C+and+Chao+Zhang%2E+Aligning+large+language+models+with+representation+editing%3A+A+control+per)
31. Sandipan Kundu, Yuntao Bai, Saurav Kadavath, Amanda Askell, Andrew Callahan, Anna Chen, Anna Goldie, Avital Balwit, Azalia Mirhoseini, Brayden McLean, et al. Specific versus general principles for constitutional ai. arXiv preprint arXiv:2310.13798, 2023. [find](https://scholar.google.com/scholar?q=Sandipan+Kundu%2C+Yuntao+Bai%2C+Saurav+Kadavath%2C+Amanda+Askell%2C+Andrew+Callahan%2C+Anna+Chen%2C+Anna+Goldie%2C+Avital+Balwit%2C+Azalia+Mirhoseini%2C+Brayden+McLean%2C+et+al%2E+Specific+versus+general+principles+for+con)
32. Maxime Labonne. Uncensor any llm with abliteration. [https://huggingface.co/blog/](https://huggingface.co/blog/) mlabonne/abliteration, 2024. [↗](https://huggingface.co/blog/)
33. Vedang Lad, Wes Gurnee, and Max Tegmark. The remarkable robustness of llms: Stages of inference? arXiv preprint arXiv:2406.19384, 2024. — [arXiv:2406.19384](https://arxiv.org/abs/2406.19384) [find](https://scholar.google.com/scholar?q=Vedang+Lad%2C+Wes+Gurnee%2C+and+Max+Tegmark%2E+The+remarkable+robustness+of+llms%3A+Stages+of+inference%3F+arXiv+preprint+arXiv%3A2406%2E19384%2C+2024%2E+%E2%80%94+%5BarXiv%3A2406%2E19384%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2406%2E19384%29)
34. Ariel N. Lee, Cole J. Hunter, and Nataniel Ruiz. [Platypus: Quick, cheap, and powerful refinement of llms](https://md.research.elamur.ai/free/en/papers/lee-2023-platypus-quick-cheap-powerful-refinement/). 2023a.
35. Bruce W Lee, Hyunsoo Cho, and Kang Min Yoo. Instruction tuning with human curriculum. arXiv preprint arXiv:2310.09518, 2023b. [find](https://scholar.google.com/scholar?q=Bruce+W+Lee%2C+Hyunsoo+Cho%2C+and+Kang+Min+Yoo%2E+Instruction+tuning+with+human+curriculum%2E+arXiv+preprint+arXiv%3A2310%2E09518%2C+2023b%2E)
36. Jingling Li, Zeyu Tang, Xiaoyu Liu, Peter Spirtes, Kun Zhang, Liu Leqi, and Yang Liu. Steering llms towards unbiased responses: A causality-guided debiasing framework. arXiv preprint arXiv:2403.08743, 2024. — [arXiv:2403.08743](https://arxiv.org/abs/2403.08743) [find](https://scholar.google.com/scholar?q=Jingling+Li%2C+Zeyu+Tang%2C+Xiaoyu+Liu%2C+Peter+Spirtes%2C+Kun+Zhang%2C+Liu+Leqi%2C+and+Yang+Liu%2E+Steering+llms+towards+unbiased+responses%3A+A+causality%2Dguided+debiasing+framework%2E+arXiv+preprint+arXiv%3A2403%2E08743%2C)
37. Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. preprint arXiv:2305.[20050, 2023](https://md.research.elamur.ai/free/en/papers/lightman-2023-let-s-verify-step-step/).
38. X Liu, Y Zhu, J Gu, Y Lan, C Yang, and Y Qiao. [Mm-safetybench: A benchmark for safety evaluation of multimodal large language models](https://md.research.elamur.ai/free/en/papers/liu-2023-mm-safetybench-benchmark-safety-evaluation/). arXiv preprint arXiv:2311.17600, 2023.
39. Ryan Louie, Ananjan Nandi, William Fang, Cheng Chang, Emma Brunskill, and Diyi Yang. [Roleplay-doh: Enabling domain-experts to create llm-simulated patients via eliciting and adhering to principles](https://md.research.elamur.ai/free/en/papers/louie-2024-roleplay-doh-enabling-domain-experts/). arXiv preprint arXiv:2407.00870, 2024.
40. Dawn Lu and Nina Rimsky. Investigating bias representations in llama 2 chat via activation steering, 2024. [find](https://scholar.google.com/scholar?q=Dawn+Lu+and+Nina+Rimsky%2E+Investigating+bias+representations+in+llama+2+chat+via+activation+steering%2C+2024%2E)
41. Pan Lu, Swaroop Mishra, Tony Xia, Liang Qiu, Kai-Wei Chang, Song-Chun Zhu, Oyvind Tafjord, Peter Clark, and Ashwin Kalyan. Learn to explain: Multimodal reasoning via thought chains for science question answering. In The 36th Conference on Neural Information Processing Systems (NeurIPS), 2022. [find](https://scholar.google.com/scholar?q=Pan+Lu%2C+Swaroop+Mishra%2C+Tony+Xia%2C+Liang+Qiu%2C+Kai%2DWei+Chang%2C+Song%2DChun+Zhu%2C+Oyvind+Tafjord%2C+Peter+Clark%2C+and+Ashwin+Kalyan%2E+Learn+to+explain%3A+Multimodal+reasoning+via+thought+chains+for+science+questio)
42. Brandon McKinzie, Zhe Gan, Jean-Philippe Fauconnier, Sam Dodge, Bowen Zhang, Philipp Dufter, Dhruti Shah, Xianzhi Du, Futang Peng, Floris Weers, et al. Mm1: Methods, analysis & insights from multimodal llm pre-training. arXiv preprint arXiv:2403.09611, 2024. — [arXiv:2403.09611](https://arxiv.org/abs/2403.09611) [find](https://scholar.google.com/scholar?q=Brandon+McKinzie%2C+Zhe+Gan%2C+Jean%2DPhilippe+Fauconnier%2C+Sam+Dodge%2C+Bowen+Zhang%2C+Philipp+Dufter%2C+Dhruti+Shah%2C+Xianzhi+Du%2C+Futang+Peng%2C+Floris+Weers%2C+et+al%2E+Mm1%3A+Methods%2C+analysis+%26+insights+from+multimoda)
43. Meta. Introducing meta llama 3: The most capable openly available llm to date. [https://ai.meta.com/blog/meta-llama-3/](https://ai.meta.com/blog/meta-llama-3/), 2024. [↗](https://ai.meta.com/blog/meta-llama-3/)
44. Katherine Metcalf, Miguel Sarabia, Natalie Mackraz, and Barry-John Theobald. Sample-efficient preference-based reinforcement learning with dynamics aware rewards. arXiv preprint arXiv:2402.17975, 2024. — [arXiv:2402.17975](https://arxiv.org/abs/2402.17975) [find](https://scholar.google.com/scholar?q=Katherine+Metcalf%2C+Miguel+Sarabia%2C+Natalie+Mackraz%2C+and+Barry%2DJohn+Theobald%2E+Sample%2Defficient+preference%2Dbased+reinforcement+learning+with+dynamics+aware+rewards%2E+arXiv+preprint+arXiv%3A2402%2E17975%2C+2024)
45. Manish Nagireddy, Lamogha Chiazor, Moninder Singh, and Ioana Baldini. Socialstigmaqa: A benchmark to uncover stigma amplification in generative language models, 2023. [find](https://scholar.google.com/scholar?q=Manish+Nagireddy%2C+Lamogha+Chiazor%2C+Moninder+Singh%2C+and+Ioana+Baldini%2E+Socialstigmaqa%3A+A+benchmark+to+uncover+stigma+amplification+in+generative+language+models%2C+2023%2E)
46. Kiho Park, Yo Joong Choe, and Victor Veitch. [The linear representation hypothesis and the geometry of large language models](https://md.research.elamur.ai/free/en/papers/park-2023-linear-representation-hypothesis-geometry-large/). arXiv preprint arXiv:2311.03658, 2023.
47. Baolin Peng, Chunyuan Li, Pengcheng He, Michel Galley, and Jianfeng Gao. [Instruction tuning with gpt-4](https://md.research.elamur.ai/free/en/papers/peng-2023-instruction-tuning-gpt-4/). arXiv preprint arXiv:2304.03277, 2023.
48. Pascal Pfeiffer, Philipp Singer, Yauhen Babakhin, Gabor Fodor, Nischay Dhankhar, and Sri Satish Ambati. H2o-danube3 technical report. arXiv preprint arXiv:2407.09276, 2024. — [arXiv:2407.09276](https://arxiv.org/abs/2407.09276) [find](https://scholar.google.com/scholar?q=Pascal+Pfeiffer%2C+Philipp+Singer%2C+Yauhen+Babakhin%2C+Gabor+Fodor%2C+Nischay+Dhankhar%2C+and+Sri+Satish+Ambati%2E+H2o%2Ddanube3+technical+report%2E+arXiv+preprint+arXiv%3A2407%2E09276%2C+2024%2E+%E2%80%94+%5BarXiv%3A2407%2E09276%5D%28https%3A)
49. Phuc Phan, Hieu Tran, and Long Phan. [Distillation contrastive decoding: Improving llms reasoning with contrastive decoding and distillation](https://md.research.elamur.ai/free/en/papers/phan-2024-distillation-contrastive-decoding-improving-llms/). arXiv preprint arXiv:2402.14874, 2024.
50. Silviu Pitis. Failure modes of learning reward models for llms and other sequence models. In ICML 2023 Workshop The Many Facets of Preference-Based Learning, 2023. [find](https://scholar.google.com/scholar?q=Silviu+Pitis%2E+Failure+modes+of+learning+reward+models+for+llms+and+other+sequence+models%2E+In+ICML+2023+Workshop+The+Many+Facets+of+Preference%2DBased+Learning%2C+2023%2E)
51. Yifu Qiu, Zheng Zhao, Yftah Ziser, Anna Korhonen, Edoardo M Ponti, and Shay B Cohen. Spectral editing of activations for large language model alignment. arXiv preprint arXiv:2405.09719, 2024. — [arXiv:2405.09719](https://arxiv.org/abs/2405.09719) [find](https://scholar.google.com/scholar?q=Yifu+Qiu%2C+Zheng+Zhao%2C+Yftah+Ziser%2C+Anna+Korhonen%2C+Edoardo+M+Ponti%2C+and+Shay+B+Cohen%2E+Spectral+editing+of+activations+for+large+language+model+alignment%2E+arXiv+preprint+arXiv%3A2405%2E09719%2C+2024%2E+%E2%80%94+%5BarXiv)
52. Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. Improving language understanding by generative pre-training. 2018. [find](https://scholar.google.com/scholar?q=Alec+Radford%2C+Karthik+Narasimhan%2C+Tim+Salimans%2C+Ilya+Sutskever%2C+et+al%2E+Improving+language+understanding+by+generative+pre%2Dtraining%2E+2018%2E)
53. Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. [Direct preference optimization: Your language model is secretly a reward model](https://md.research.elamur.ai/free/en/papers/rafailov-2023-direct-preference-optimization-language-model/). Advances in Neural Information Processing Systems, 36, 2024.
54. Nate Rahn, Pierluca D’Oro, and Marc G Bellemare. Controlling large language model agents with entropic activation steering. arXiv preprint arXiv:2406.00244, 2024. — [arXiv:2406.00244](https://arxiv.org/abs/2406.00244) [find](https://scholar.google.com/scholar?q=Nate+Rahn%2C+Pierluca+D%E2%80%99Oro%2C+and+Marc+G+Bellemare%2E+Controlling+large+language+model+agents+with+entropic+activation+steering%2E+arXiv+preprint+arXiv%3A2406%2E00244%2C+2024%2E+%E2%80%94+%5BarXiv%3A2406%2E00244%5D%28https%3A%2F%2Farxiv%2Eor)
55. Max Reuter and William Schulze. I’m afraid i can’t do that: Predicting prompt refusal in black-box generative language models. arXiv preprint arXiv:2306.03423, 2023. — [arXiv:2306.03423](https://arxiv.org/abs/2306.03423) [find](https://scholar.google.com/scholar?q=Max+Reuter+and+William+Schulze%2E+I%E2%80%99m+afraid+i+can%E2%80%99t+do+that%3A+Predicting+prompt+refusal+in+black%2Dbox+generative+language+models%2E+arXiv+preprint+arXiv%3A2306%2E03423%2C+2023%2E+%E2%80%94+%5BarXiv%3A2306%2E03423%5D%28https%3A%2F%2Farxiv)
56. Nina Rimsky, Nick Gabrieli, Julian Schulz, Meg Tong, Evan Hubinger, and Alexander Matt Turner. Steering llama 2 via contrastive activation addition, 2024. [find](https://scholar.google.com/scholar?q=Nina+Rimsky%2C+Nick+Gabrieli%2C+Julian+Schulz%2C+Meg+Tong%2C+Evan+Hubinger%2C+and+Alexander+Matt+Turner%2E+Steering+llama+2+via+contrastive+activation+addition%2C+2024%2E)
57. Shibani Santurkar, Esin Durmus, Faisal Ladhak, Cinoo Lee, Percy Liang, and Tatsunori Hashimoto. [Whose opinions do language models reflect?](https://md.research.elamur.ai/free/en/papers/santurkar-2023-whose-opinions-language-models-reflect/) In International Conference on Machine Learning, pp. 29971–30004. PMLR, 2023.
58. Tomohiro Sawada, Daniel Paleka, Alexander Havrilla, Pranav Tadepalli, Paula Vidas, Alexander Kranias, John J. Nay, Kshitij Gupta, and Aran Komatsuzaki. [Arb: Advanced reasoning benchmark for large language models](https://md.research.elamur.ai/free/en/papers/sawada-2023-arb-advanced-reasoning-benchmark-large/), 2023.
59. Daniel Scalena, Gabriele Sarti, and Malvina Nissim. Multi-property steering of large language models with dynamic activation composition. arXiv preprint arXiv:2406.17563, 2024. [find](https://scholar.google.com/scholar?q=Daniel+Scalena%2C+Gabriele+Sarti%2C+and+Malvina+Nissim%2E+Multi%2Dproperty+steering+of+large+language+models+with+dynamic+activation+composition%2E+arXiv+preprint+arXiv%3A2406%2E17563%2C+2024%2E)
60. Adam S Shai, Sarah E Marzen, Lucas Teixeira, Alexander Gietelink Oldenziel, and Paul M Riechers. Transformers represent belief state geometry in their residual stream. arXiv preprint arXiv:2405.15943, 2024. — [arXiv:2405.15943](https://arxiv.org/abs/2405.15943) [find](https://scholar.google.com/scholar?q=Adam+S+Shai%2C+Sarah+E+Marzen%2C+Lucas+Teixeira%2C+Alexander+Gietelink+Oldenziel%2C+and+Paul+M+Riechers%2E+Transformers+represent+belief+state+geometry+in+their+residual+stream%2E+arXiv+preprint+arXiv%3A2405%2E15943%2C)
61. Taylor Sorensen, Jared Moore, Jillian Fisher, Mitchell Gordon, Niloofar Mireshghallah, Christopher Michael Rytting, Andre Ye, Liwei Jiang, Ximing Lu, Nouha Dziri, et al. [A roadmap to pluralistic alignment](https://md.research.elamur.ai/free/en/papers/sorensen-2024-roadmap-pluralistic-alignment/). arXiv preprint arXiv:2402.05070, 2024.
62. Ondrej Sotolar. Empo: Theory-driven dataset construction for empathetic response generation through preference optimization. arXiv preprint arXiv:2406.19071, 2024. — [arXiv:2406.19071](https://arxiv.org/abs/2406.19071) [find](https://scholar.google.com/scholar?q=Ondrej+Sotolar%2E+Empo%3A+Theory%2Ddriven+dataset+construction+for+empathetic+response+generation+through+preference+optimization%2E+arXiv+preprint+arXiv%3A2406%2E19071%2C+2024%2E+%E2%80%94+%5BarXiv%3A2406%2E19071%5D%28https%3A%2F%2Farxiv%2Eo)
63. Asa Cooper Stickland, Alexander Lyzhov, Jacob Pfau, Salsabila Mahdi, and Samuel R Bowman. Steering without side effects: Improving post-deployment control of language models. arXiv preprint arXiv:2406.15518, 2024. [find](https://scholar.google.com/scholar?q=Asa+Cooper+Stickland%2C+Alexander+Lyzhov%2C+Jacob+Pfau%2C+Salsabila+Mahdi%2C+and+Samuel+R+Bowman%2E+Steering+without+side+effects%3A+Improving+post%2Ddeployment+control+of+language+models%2E+arXiv+preprint+arXiv%3A2406)
64. Nisan Stiennon, Long Ouyang, Jeffrey Wu, Daniel Ziegler, Ryan Lowe, Chelsea Voss, Alec Radford, Dario Amodei, and Paul F Christiano. Learning to summarize with human feedback. Advances in Neural Information Processing Systems, 33:3008–3021, 2020. [find](https://scholar.google.com/scholar?q=Nisan+Stiennon%2C+Long+Ouyang%2C+Jeffrey+Wu%2C+Daniel+Ziegler%2C+Ryan+Lowe%2C+Chelsea+Voss%2C+Alec+Radford%2C+Dario+Amodei%2C+and+Paul+F+Christiano%2E+Learning+to+summarize+with+human+feedback%2E+Advances+in+Neural+Infor)
65. Shivchander Sudalairaj, Abhishek Bhandwaldar, Aldo Pareja, Kai Xu, David D Cox, and Akash Srivastava. Lab: Large-scale alignment for chatbots. arXiv preprint arXiv:2403.01081, 2024. — [arXiv:2403.01081](https://arxiv.org/abs/2403.01081) [find](https://scholar.google.com/scholar?q=Shivchander+Sudalairaj%2C+Abhishek+Bhandwaldar%2C+Aldo+Pareja%2C+Kai+Xu%2C+David+D+Cox%2C+and+Akash+Srivastava%2E+Lab%3A+Large%2Dscale+alignment+for+chatbots%2E+arXiv+preprint+arXiv%3A2403%2E01081%2C+2024%2E+%E2%80%94+%5BarXiv%3A2403%2E0108)
66. Hovhannes Tamoyan, Hendrik Schuff, and Iryna Gurevych. Llm roleplay: Simulating human-chatbot interaction. arXiv preprint arXiv:2407.03974, 2024. [find](https://scholar.google.com/scholar?q=Hovhannes+Tamoyan%2C+Hendrik+Schuff%2C+and+Iryna+Gurevych%2E+Llm+roleplay%3A+Simulating+human%2Dchatbot+interaction%2E+arXiv+preprint+arXiv%3A2407%2E03974%2C+2024%2E)
67. Daniel Chee Hian Tan, David Chanin, Aengus Lynch, Adrià Garriga-Alonso, Dimitrios Kanoulas, Brooks Paige, and Robert Kirk. Analyzing the generalization and reliability of steering vectors. In ICML 2024 Workshop on Mechanistic Interpretability, 2024. [find](https://scholar.google.com/scholar?q=Daniel+Chee+Hian+Tan%2C+David+Chanin%2C+Aengus+Lynch%2C+Adri%C3%A0+Garriga%2DAlonso%2C+Dimitrios+Kanoulas%2C+Brooks+Paige%2C+and+Robert+Kirk%2E+Analyzing+the+generalization+and+reliability+of+steering+vectors%2E+In+ICML+202)
68. Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B Hashimoto. Stanford alpaca: An instruction-following llama model, 2023. [find](https://scholar.google.com/scholar?q=Rohan+Taori%2C+Ishaan+Gulrajani%2C+Tianyi+Zhang%2C+Yann+Dubois%2C+Xuechen+Li%2C+Carlos+Guestrin%2C+Percy+Liang%2C+and+Tatsunori+B+Hashimoto%2E+Stanford+alpaca%3A+An+instruction%2Dfollowing+llama+model%2C+2023%2E)
69. Yi Tay, Jason Wei, Hyung Won Chung, Vinh Q Tran, David R So, Siamak Shakeri, Xavier Garcia, Huaixiu Steven Zheng, Jinfeng Rao, Aakanksha Chowdhery, et al. [Transcending scaling laws with 0.1% extra compute](https://md.research.elamur.ai/free/en/papers/tay-2022-transcending-scaling-laws-0-1/). arXiv preprint arXiv:2210.11399, 2022.
70. Teknium, interstellarninja, theemozilla, karan4d, and huemin_art. Hermes-2-pro-llama-3-8b, 2024. URL [https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B](https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B). [↗](https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B)
71. Alejandro Tlaie. Exploring and steering the moral compass of large language models. arXiv preprint arXiv:2405.17345, 2024. — [arXiv:2405.17345](https://arxiv.org/abs/2405.17345) [find](https://scholar.google.com/scholar?q=Alejandro+Tlaie%2E+Exploring+and+steering+the+moral+compass+of+large+language+models%2E+arXiv+preprint+arXiv%3A2405%2E17345%2C+2024%2E+%E2%80%94+%5BarXiv%3A2405%2E17345%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2405%2E17345%29)
72. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. [Llama 2: Open foundation and fine-tuned chat models](https://md.research.elamur.ai/free/en/papers/touvron-2023-llama-2-open-foundation-fine/). arXiv preprint arXiv:2307.09288, 2023.
73. Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sanseviero, Alexander M. Rush, and Thomas Wolf. Zephyr: Direct distillation of lm alignment, 2023. [find](https://scholar.google.com/scholar?q=Lewis+Tunstall%2C+Edward+Beeching%2C+Nathan+Lambert%2C+Nazneen+Rajani%2C+Kashif+Rasul%2C+Younes+Belkada%2C+Shengyi+Huang%2C+Leandro+von+Werra%2C+Cl%C3%A9mentine+Fourrier%2C+Nathan+Habib%2C+Nathan+Sarrazin%2C+Omar+Sanseviero%2C+Al)
74. Alex Turner, Lisa Thiergart, David Udell, Gavin Leech, Ulisse Mini, and Monte MacDiarmid. Activation addition: Steering language models without optimization. arXiv preprint arXiv:2308.[10248, 2023](https://md.research.elamur.ai/free/en/papers/turner-2023-steering-language-models-activation-engineering/).
75. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. [Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/). Advances in neural information processing systems, 30, 2017.
76. Tianlong Wang, Xianfeng Jiao, Yifan He, Zhongzhi Chen, Yinghao Zhu, Xu Chu, Junyi Gao, Yasha Wang, and Liantao Ma. Adaptive activation steering: A tuning-free llm truthfulness improvement method for diverse hallucinations categories. arXiv preprint arXiv:2406.00034, 2024a. — [arXiv:2406.00034](https://arxiv.org/abs/2406.00034) [find](https://scholar.google.com/scholar?q=Tianlong+Wang%2C+Xianfeng+Jiao%2C+Yifan+He%2C+Zhongzhi+Chen%2C+Yinghao+Zhu%2C+Xu+Chu%2C+Junyi+Gao%2C+Yasha+Wang%2C+and+Liantao+Ma%2E+Adaptive+activation+steering%3A+A+tuning%2Dfree+llm+truthfulness+improvement+method+for+d)
77. Xiaoxuan Wang, Ziniu Hu, Pan Lu, Yanqiao Zhu, Jieyu Zhang, Satyen Subramaniam, Arjun R. Loomba, Shichang Zhang, Yizhou Sun, and Wei Wang. [Scibench: Evaluating college-level scientific problem-solving abilities of large language models](https://md.research.elamur.ai/free/en/papers/wang-2023-scibench-evaluating-college-level-scientific/), 2023a.
78. Yuxia Wang, Haonan Li, Xudong Han, Preslav Nakov, and Timothy Baldwin. [Do-not-answer: A dataset for evaluating safeguards in llms](https://md.research.elamur.ai/free/en/papers/wang-2023-answer-dataset-evaluating-safeguards-llms/). arXiv preprint arXiv:2308.13387, 2023b.
79. Zihao Wang and Victor Veitch. Does editing provide evidence for localization? In ICML 2024 Workshop on Mechanistic Interpretability. [find](https://scholar.google.com/scholar?q=Zihao+Wang+and+Victor+Veitch%2E+Does+editing+provide+evidence+for+localization%3F+In+ICML+2024+Workshop+on+Mechanistic+Interpretability%2E)
80. Zihao Wang, Lin Gui, Jeffrey Negrea, and Victor Veitch. Concept algebra for (score-based) text-controlled generative models. Advances in Neural Information Processing Systems, 36, 2024b. [find](https://scholar.google.com/scholar?q=Zihao+Wang%2C+Lin+Gui%2C+Jeffrey+Negrea%2C+and+Victor+Veitch%2E+Concept+algebra+for+%28score%2Dbased%29+text%2Dcontrolled+generative+models%2E+Advances+in+Neural+Information+Processing+Systems%2C+36%2C+2024b%2E)
81. Bingbing Wen, Jihan Yao, Shangbin Feng, Chenjun Xu, Yulia Tsvetkov, Bill Howe, and Lucy Lu Wang. The art of refusal: A survey of abstention in large language models. arXiv preprint arXiv:2407.18418, 2024. [find](https://scholar.google.com/scholar?q=Bingbing+Wen%2C+Jihan+Yao%2C+Shangbin+Feng%2C+Chenjun+Xu%2C+Yulia+Tsvetkov%2C+Bill+Howe%2C+and+Lucy+Lu+Wang%2E+The+art+of+refusal%3A+A+survey+of+abstention+in+large+language+models%2E+arXiv+preprint+arXiv%3A2407%2E18418%2C+2)
82. Zhengxuan Wu, Aryaman Arora, Zheng Wang, Atticus Geiger, Dan Jurafsky, Christopher D Manning, and Christopher Potts. [Reft: Representation finetuning for language models](https://md.research.elamur.ai/free/en/papers/wu-2024-reft-representation-finetuning-language-models/). arXiv preprint arXiv:2404.03592, 2024.
83. Qianqian Xie, Qingyu Chen, Aokun Chen, Cheng Peng, Yan Hu, Fongci Lin, Xueqing Peng, Jimin Huang, Jeffrey Zhang, Vipina Keloth, et al. [Me llama: Foundation large language models for medical applications](https://md.research.elamur.ai/free/en/papers/xie-2024-me-llama-foundation-large-language/). arXiv preprint arXiv:2402.12749, 2024a.
84. Tinghao Xie, Xiangyu Qi, Yi Zeng, Yangsibo Huang, Udari Madhushani Sehwag, Kaixuan Huang, Luxi He, Boyi Wei, Dacheng Li, Ying Sheng, et al. [Sorry-bench: Systematically evaluating large language model safety refusal](https://md.research.elamur.ai/free/en/papers/xie-2024-sorry-bench-systematically-evaluating-large/) behaviors. arXiv preprint arXiv:2406.14598, 2024b.
85. Can Xu, Qingfeng Sun, Kai Zheng, Xiubo Geng, Pu Zhao, Jiazhan Feng, Chongyang Tao, Qingwei Lin, and Daxin Jiang. [Wizardlm: Empowering large pre-trained language models to follow complex instructions](https://md.research.elamur.ai/free/en/papers/xu-2023-wizardlm-empowering-large-pre-trained/). In The Twelfth International Conference on Learning Representations, 2024.
86. Fangcong Yin, Xi Ye, and Greg Durrett. Lofit: Localized fine-tuning on llm representations. arXiv preprint arXiv:2406.01563, 2024. — [arXiv:2406.01563](https://arxiv.org/abs/2406.01563) [find](https://scholar.google.com/scholar?q=Fangcong+Yin%2C+Xi+Ye%2C+and+Greg+Durrett%2E+Lofit%3A+Localized+fine%2Dtuning+on+llm+representations%2E+arXiv+preprint+arXiv%3A2406%2E01563%2C+2024%2E+%E2%80%94+%5BarXiv%3A2406%2E01563%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2406%2E01563%29)
87. Weihao Yu, Zihang Jiang, Yanfei Dong, and Jiashi Feng. [Reclor: A reading comprehension dataset requiring logical reasoning](https://md.research.elamur.ai/free/en/papers/yu-2020-reclor-reading-comprehension-dataset-requiring/). In International Conference on Learning Representations (ICLR), April 2020.
88. Jie Zhang, Dongrui Liu, Chen Qian, Ziyue Gan, Yong Liu, Yu Qiao, and Jing Shao. The better angels of machine personality: How personality relates to llm safety. arXiv preprint arXiv:2407.12344, 2024. — [arXiv:2407.12344](https://arxiv.org/abs/2407.12344) [find](https://scholar.google.com/scholar?q=Jie+Zhang%2C+Dongrui+Liu%2C+Chen+Qian%2C+Ziyue+Gan%2C+Yong+Liu%2C+Yu+Qiao%2C+and+Jing+Shao%2E+The+better+angels+of+machine+personality%3A+How+personality+relates+to+llm+safety%2E+arXiv+preprint+arXiv%3A2407%2E12344%2C+2024%2E+)
89. Chujie Zheng, Fan Yin, Hao Zhou, Fandong Meng, Jie Zhou, Kai-Wei Chang, Minlie Huang, and Nanyun Peng. Prompt-driven llm safeguarding via directed representation optimization. arXiv preprint arXiv:2401.[18018, 2024](https://md.research.elamur.ai/free/en/papers/zheng-2024-prompt-driven-safeguarding-large-language/).
90. Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Tianle Li, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zhuohan Li, Zi Lin, Eric. P Xing, Joseph E. Gonzalez, Ion Stoica, and Hao Zhang. [Lmsys-chat-1m: A large-scale real-world llm conversation dataset](https://md.research.elamur.ai/free/en/papers/zheng-2023-lmsys-chat-1m-large-scale/), 2023.
91. Andy Zou, Long Phan, Sarah Chen, James Campbell, Phillip Guo, Richard Ren, Alexander Pan, Xuwang Yin, Mantas Mazeika, Ann-Kathrin Dombrowski, Shashwat Goel, Nathaniel Li, Michael J. Byun, Zifan Wang, Alex Mallen, Steven Basart, Sanmi Koyejo, Dawn Song, Matt Fredrikson, J. Zico Kolter, and Dan Hendry… — [Representation Engineering: A Top-Down Approach to AI Transparency](https://md.research.elamur.ai/free/en/papers/zou-2023-representation-engineering-top-down-approach/)

*Updated 3 November 2026*