---
title: "Deciphering Cultural Representations in Large Language Models"
description: "They introduce a selectivity score that ranks features by how differently they activate across cultures. 59 references, cited by 1 paper here."
authors: ["Chenye Zou", "Difan Jiao", "Lijie Hu"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.278/"
doi: "10.18653/v1/2026.findings-acl.278"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zou-2026-deciphering-cultural-representations-large-language/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zou-2026-deciphering-cultural-representations-large-language/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/11/) / Deciphering Cultural Representations in Large Language Models via Sparse Autoencoders

# Deciphering Cultural Representations in Large Language Models via Sparse Autoencoders

**Authors:** Chenye Zou, Difan Jiao and [Lijie Hu](https://md.research.elamur.ai/free/en/authors/lijie-hu/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.278/) · [PDF](https://aclanthology.org/2026.findings-acl.278.pdf) · [DOI 10.18653/v1/2026.findings-acl.278](https://doi.org/10.18653/v1/2026.findings-acl.278) · [OpenAlex](https://openalex.org/W7166876200)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
LLMs tend to favor Western cultural defaults, and it is unclear how they store cultural knowledge internally. Earlier "cultural neuron" studies are limited because single neurons often mix several unrelated concepts. This work uses pretrained Sparse Autoencoders (Gemma Scope for Gemma-2-9B, Llama Scope for Llama-3.1-8B) to split residual-stream activations into sparse, more interpretable features. The authors evaluate on a combined cultural question set called CPC, built from WVS, the Pew Global Attitudes Survey (GAS), NormAd and BLEnD.

They introduce a selectivity score that ranks features by how differently they activate across cultures. The top-ranked features respond consistently across rephrasings and task formats. Removing these features shows they are needed for culture-dependent answers. Amplifying them shows they can raise cultural accuracy without any retraining.

The main takeaway is that cultural knowledge appears to be encoded abstractly, not only through surface word cues. This makes feature-level intervention a cheap, interpretable option for cultural alignment, including for under-resourced cultures.

## Key points
- **Data:** CPC contains about 1,500 questions per country. Experiments cover eight cultures, coded AS, CN, ES, GR, ID, IR, NG and US.
- **Layer-wise finding:** In upper layers, representations of the same culture become more similar across tasks than representations of different cultures within a task. In early layers, task format dominates instead.
- **Ablation:** Suppressing only the top-10 culture features in layers 22–26 drops accuracy from 0.57 to 0.31. Ablating a matched set of random features barely changes performance.
- **Steering results (average accuracy):**
  - Llama-3.1-8B: 0.58 zero-shot, 0.63 with culture prompting, 0.66 with SAE steering.
  - Gemma-2-9B: 0.56 zero-shot, 0.60 with culture prompting, 0.64 with SAE steering.
  - Combining prompting with steering gave no consistent extra gain.
- **Efficiency and limitations:** Finding steering directions needs only about 50 labeled examples per culture. The study is limited to English-language data and survey-based proxies for culture. The steering strength must be tuned separately for each culture.

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

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🤝 Safety Research](https://md.research.elamur.ai/free/en/shelves/social-sciences/safety-research/) · 543 works

🧭 Topics (1): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,320 works

## Cite

Cite as: Chenye Zou, Difan Jiao and Lijie Hu. 2026. Deciphering Cultural Representations in Large Language Models via Sparse Autoencoders. Findings of the Association for Computational Linguistics: ACL 2026.

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

```bibtex
@inproceedings{zou2026deciphering,
  title = {{Deciphering Cultural Representations in Large Language Models via Sparse Autoencoders}},
  author = {Chenye Zou, Difan Jiao and Lijie Hu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.278},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.278}
}
```

- [← previous Decentralized Arena: Towards Democratic and Scalable Automatic Evaluation of Language Models](https://md.research.elamur.ai/free/en/papers/yin-2026-decentralized-arena-democratic-scalable-automatic/)
- [next → Decoding the Multimodal Mind: Generalizable Brain-to-Text Translation via Multimodal Alignment and Adaptive Routing](https://md.research.elamur.ai/free/en/papers/ye-2026-decoding-multimodal-mind-generalizable-brain/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/11/)

## References

59 references, 16 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Badr AlKhamissi, Muhammad ElNokrashy, Mai Alkhamissi, and Mona Diab. 2024. [Investigating cultural alignment of large language models](https://md.research.elamur.ai/free/en/papers/alkhamissi-2024-investigating-cultural-alignment-large-language/). In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12404–12422, Bangkok, Thailand. Association…
2. Andy Arditi, Oscar Obeso, Aaquib Syed, Daniel Paleka, Nina Panickssery, Wes Gurnee, and Neel Nanda. 2024. [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/). In Proceedings of the 38th International Conference on Neural Information Processing Systems, NIPS ’24, Red Hook, NY, USA. Curran As…
3. Trenton Bricken, Adly Templeton, Joshua Batson, Brian Chen, Adam Jermyn, Tom Conerly, Nick Turner, Cem Anil, Carson Denison, Amanda Askell, Robert Lasenby, Yifan Wu, Shauna Kravec, Nicholas Schiefer, Tim Maxwell, Nicholas Joseph, Zac Hatfield-Dodds, Alex Tamkin, Karina Nguyen, Brayden McLean, Josiah… [find](https://scholar.google.com/scholar?q=Trenton+Bricken%2C+Adly+Templeton%2C+Joshua+Batson%2C+Brian+Chen%2C+Adam+Jermyn%2C+Tom+Conerly%2C+Nick+Turner%2C+Cem+Anil%2C+Carson+Denison%2C+Amanda+Askell%2C+Robert+Lasenby%2C+Yifan+Wu%2C+Shauna+Kravec%2C+Nicholas+Schiefer%2C+)
4. Keyuan Cheng, Zijian Kan, Zhuoran Zhang, Muhammad Asif Ali, Lijie Hu, and Di Wang. 2025a. Compke: Complex question answering under knowledge editing. In Findings of the Association for Computational Linguistics: ACL 2025, pages 2557–2576. [find](https://scholar.google.com/scholar?q=Keyuan+Cheng%2C+Zijian+Kan%2C+Zhuoran+Zhang%2C+Muhammad+Asif+Ali%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025a%2E+Compke%3A+Complex+question+answering+under+knowledge+editing%2E+In+Findings+of+the+Association+for+Computational+L)
5. Mehar Bhatia, Maria Antoniak, Yulia Tsvetkov, Vered Shwartz, and Yejin Choi. 2025. [CulturalBench: A robust, diverse and challenging benchmark for measuring LMs’ cultural knowledge through human-AI red-teaming](https://md.research.elamur.ai/free/en/papers/chiu-2025-culturalbench-robust-diverse-challenging-benchmark/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics…
6. Hoagy Cunningham, Aidan Ewart, Logan Riggs, Robert Huben, and Lee Sharkey. 2023. Sparse autoencoders find highly interpretable features in language models. Preprint, arXiv:2309.08600. [find](https://scholar.google.com/scholar?q=Hoagy+Cunningham%2C+Aidan+Ewart%2C+Logan+Riggs%2C+Robert+Huben%2C+and+Lee+Sharkey%2E+2023%2E+Sparse+autoencoders+find+highly+interpretable+features+in+language+models%2E+Preprint%2C+arXiv%3A2309%2E08600%2E)
7. Boyi Deng, Yu Wan, Baosong Yang, Yidan Zhang, and Fuli Feng. 2025. [Unveiling language-specific features in large language models via sparse autoencoders](https://md.research.elamur.ai/free/en/papers/deng-2025-unveiling-language-specific-features-large/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4563–4608, Vienna, Austr…
8. Jesse Dodge, Maarten Sap, Ana Marasović, William Agnew, Gabriel Ilharco, Dirk Groeneveld, Margaret Mitchell, and Matt Gardner. 2021. Documenting large webtext corpora: A case study on the colossal clean crawled corpus. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language P… [find](https://scholar.google.com/scholar?q=Jesse+Dodge%2C+Maarten+Sap%2C+Ana+Marasovic%CC%81%2C+William+Agnew%2C+Gabriel+Ilharco%2C+Dirk+Groeneveld%2C+Margaret+Mitchell%2C+and+Matt+Gardner%2E+2021%2E+Documenting+large+webtext+corpora%3A+A+case+study+on+the+colossal+cl)
9. Nelson Elhage, Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Ja… [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+Nova+DasSarma%2C+Dawn+Drain%2C+Deep+Ganguli%2C+Zac+Hatfield%2DDodds%2C+Da)
10. Javier Ferrando, Oscar Obeso, Senthooran Rajamanoharan, and Neel Nanda. 2025. Do i know this entity? knowledge awareness and hallucinations in language models. Preprint, arXiv:2411.14257. — [arXiv:2411.14257](https://arxiv.org/abs/2411.14257) [find](https://scholar.google.com/scholar?q=Javier+Ferrando%2C+Oscar+Obeso%2C+Senthooran+Rajamanoharan%2C+and+Neel+Nanda%2E+2025%2E+Do+i+know+this+entity%3F+knowledge+awareness+and+hallucinations+in+language+models%2E+Preprint%2C+arXiv%3A2411%2E14257%2E+%E2%80%94+%5BarXiv%3A241)
11. Keyuan Cheng, Xudong Shen, Yihao Yang, Tengyue-Wang TengyueWang, Yang Cao, Muhammad Asif Ali, Hanbin Wang, Lijie Hu, and Di Wang. 2025b. Codemenv: Benchmarking large language models on code migration. In Findings of the Association for Computational Linguistics: ACL 2025, pages 2719– 2744. [find](https://scholar.google.com/scholar?q=Keyuan+Cheng%2C+Xudong+Shen%2C+Yihao+Yang%2C+Tengyue%2DWang+TengyueWang%2C+Yang+Cao%2C+Muhammad+Asif+Ali%2C+Hanbin+Wang%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025b%2E+Codemenv%3A+Benchmarking+large+language+models+on+code+migration%2E)
12. Isabel O. Gallegos, Ryan Aponte, Ryan A. Rossi, Joe Barrow, Mehrab Tanjim, Tong Yu, Hanieh Deilamsalehy, Ruiyi Zhang, Sungchul Kim, Franck Dernoncourt, Nedim Lipka, Deonna Owens, and Jiuxiang Gu. 2025. Self-debiasing large language models: Zero-shot recognition and reduction of stereotypes. In Proce… [find](https://scholar.google.com/scholar?q=Isabel+O%2E+Gallegos%2C+Ryan+Aponte%2C+Ryan+A%2E+Rossi%2C+Joe+Barrow%2C+Mehrab+Tanjim%2C+Tong+Yu%2C+Hanieh+Deilamsalehy%2C+Ruiyi+Zhang%2C+Sungchul+Kim%2C+Franck+Dernoncourt%2C+Nedim+Lipka%2C+Deonna+Owens%2C+and+Jiuxiang+Gu%2E+2025)
13. Yu Ying Chiu, Liwei Jiang, Bill Yuchen Lin, Chan Young Park, Shuyue Stella Li, Sahithya Ravi, Atticus Geiger, Duligur Ibeling, Amir Zur, Maheep Chaudhary, Sonakshi Chauhan, Jing Huang, Arya-Keyuan Cheng, Gang Lin, Haoyang Fei, Lu Yu, Muhammad Asif Ali, Lijie Hu, Di Wang, et al. 2024. Multi-hop quest… [find](https://scholar.google.com/scholar?q=Yu+Ying+Chiu%2C+Liwei+Jiang%2C+Bill+Yuchen+Lin%2C+Chan+Young+Park%2C+Shuyue+Stella+Li%2C+Sahithya+Ravi%2C+Atticus+Geiger%2C+Duligur+Ibeling%2C+Amir+Zur%2C+Maheep+Chaudhary%2C+Sonakshi+Chauhan%2C+Jing+Huang%2C+Arya%2DKeyuan+Che)
14. Zhengfu He, Wentao Shu, Xuyang Ge, Lingjie Chen, Junxuan Wang, Yunhua Zhou, Frances Liu, Qipeng Guo, Xuanjing Huang, Zuxuan Wu, Yu-Gang Jiang, and Xipeng Qiu. 2024. Llama scope: Extracting millions of features from llama-3.1-8b with sparse autoencoders. Preprint, arXiv:2410.20526. [find](https://scholar.google.com/scholar?q=Zhengfu+He%2C+Wentao+Shu%2C+Xuyang+Ge%2C+Lingjie+Chen%2C+Junxuan+Wang%2C+Yunhua+Zhou%2C+Frances+Liu%2C+Qipeng+Guo%2C+Xuanjing+Huang%2C+Zuxuan+Wu%2C+Yu%2DGang+Jiang%2C+and+Xipeng+Qiu%2E+2024%2E+Llama+scope%3A+Extracting+millions+of)
15. Robert Huben, Hoagy Cunningham, Logan Smith, Aidan Ewart, and Lee Sharkey. 2024. Sparse autoencoders find highly interpretable features in language models. In International Conference on Representation Learning, volume 2024, pages 7827–7845. [find](https://scholar.google.com/scholar?q=Robert+Huben%2C+Hoagy+Cunningham%2C+Logan+Smith%2C+Aidan+Ewart%2C+and+Lee+Sharkey%2E+2024%2E+Sparse+autoencoders+find+highly+interpretable+features+in+language+models%2E+In+International+Conference+on+Representatio)
16. Xinyan Jiang, Lin Zhang, Jiayi Zhang, Qingsong Yang, Guimin Hu, Di Wang, and Lijie Hu. 2025. Msrs: Adaptive multi-subspace representation steering for attribute alignment in large language models. arXiv preprint arXiv:2508.10599. — [arXiv:2508.10599](https://arxiv.org/abs/2508.10599) [find](https://scholar.google.com/scholar?q=Xinyan+Jiang%2C+Lin+Zhang%2C+Jiayi+Zhang%2C+Qingsong+Yang%2C+Guimin+Hu%2C+Di+Wang%2C+and+Lijie+Hu%2E+2025%2E+Msrs%3A+Adaptive+multi%2Dsubspace+representation+steering+for+attribute+alignment+in+large+language+models%2E+arX)
17. Difan Jiao, Di Wang, and Lijie Hu. 2026. Understanding the dynamics of demonstration conflict in in-context learning. arXiv preprint arXiv:2603.04464. — [arXiv:2603.04464](https://arxiv.org/abs/2603.04464) [find](https://scholar.google.com/scholar?q=Difan+Jiao%2C+Di+Wang%2C+and+Lijie+Hu%2E+2026%2E+Understanding+the+dynamics+of+demonstration+conflict+in+in%2Dcontext+learning%2E+arXiv+preprint+arXiv%3A2603%2E04464%2E+%E2%80%94+%5BarXiv%3A2603%2E04464%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2603%2E0)
18. Hongji Li, Manjiang Yu, Priyanka Singh, Xue Li, Di Wang, Lijie Hu, et al. 2025a. Towards reasoning-preserving unlearning in multimodal large language models. arXiv preprint arXiv:2512.17911. — [arXiv:2512.17911](https://arxiv.org/abs/2512.17911) [find](https://scholar.google.com/scholar?q=Hongji+Li%2C+Manjiang+Yu%2C+Priyanka+Singh%2C+Xue+Li%2C+Di+Wang%2C+Lijie+Hu%2C+et+al%2E+2025a%2E+Towards+reasoning%2Dpreserving+unlearning+in+multimodal+large+language+models%2E+arXiv+preprint+arXiv%3A2512%2E17911%2E+%E2%80%94+%5BarXiv%3A)
19. Tong Li, Shu Yang, Junchao Wu, Jiyao Wei, Lijie Hu, Mengdi Li, Derek F. Wong, Joshua R. Oltmanns, and Di Wang. 2025b. Can large language models identify implicit suicidal ideation? an empirical evaluation. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 18392–18413, S… [find](https://scholar.google.com/scholar?q=Tong+Li%2C+Shu+Yang%2C+Junchao+Wu%2C+Jiyao+Wei%2C+Lijie+Hu%2C+Mengdi+Li%2C+Derek+F%2E+Wong%2C+Joshua+R%2E+Oltmanns%2C+and+Di+Wang%2E+2025b%2E+Can+large+language+models+identify+implicit+suicidal+ideation%3F+an+empirical+evalua)
20. Tom Lieberum, Senthooran Rajamanoharan, Arthur Conmy, Lewis Smith, Nicolas Sonnerat, Vikrant Varma, János Kramár, Anca Dragan, Rohin Shah, and Neel Nanda. 2024. Gemma scope: Open sparse autoencoders everywhere all at once on gemma 2. Preprint, arXiv:2408.05147. [find](https://scholar.google.com/scholar?q=Tom+Lieberum%2C+Senthooran+Rajamanoharan%2C+Arthur+Conmy%2C+Lewis+Smith%2C+Nicolas+Sonnerat%2C+Vikrant+Varma%2C+J%C3%A1nos+Kram%C3%A1r%2C+Anca+Dragan%2C+Rohin+Shah%2C+and+Neel+Nanda%2E+2024%2E+Gemma+scope%3A+Open+sparse+autoencoders+e)
21. Zichao Lin, Shuyan Guan, Wending Zhang, Huiyan Zhang, Yugang Li, and Huaping Zhang. 2024. Towards trustworthy llms: a review on debiasing and dehallucinating in large language models. Artificial Intelligence Review, 57. [find](https://scholar.google.com/scholar?q=Zichao+Lin%2C+Shuyan+Guan%2C+Wending+Zhang%2C+Huiyan+Zhang%2C+Yugang+Li%2C+and+Huaping+Zhang%2E+2024%2E+Towards+trustworthy+llms%3A+a+review+on+debiasing+and+dehallucinating+in+large+language+models%2E+Artificial+Intel)
22. Aleksandar Makelov, Georg Lange, Atticus Geiger, and Neel Nanda. 2024. Is this the subspace you are looking for? an interpretability illusion for subspace activation patching. In International Conference on Representation Learning, volume 2024, pages 26486– 26515. [find](https://scholar.google.com/scholar?q=Aleksandar+Makelov%2C+Georg+Lange%2C+Atticus+Geiger%2C+and+Neel+Nanda%2E+2024%2E+Is+this+the+subspace+you+are+looking+for%3F+an+interpretability+illusion+for+subspace+activation+patching%2E+In+International+Confere)
23. Reem Masoud, Ziquan Liu, Martin Ferianc, Philip C. Treleaven, and Miguel Rodrigues Rodrigues. 2025. [Cultural alignment in large language models: An explanatory analysis based on hofstede’s cultural dimensions](https://md.research.elamur.ai/free/en/papers/masoud-2025-cultural-alignment-large-language-models/). In Proceedings of the 31st International Conference on Computational Linguistics, pages 84…
24. Kevin Meng, David Bau, Alex Andonian, and Yonatan Belinkov. 2022. Locating and editing factual associations in GPT. Advances in Neural Information Processing Systems, 36. ArXiv:2202.05262. [find](https://scholar.google.com/scholar?q=Kevin+Meng%2C+David+Bau%2C+Alex+Andonian%2C+and+Yonatan+Belinkov%2E+2022%2E+Locating+and+editing+factual+associations+in+GPT%2E+Advances+in+Neural+Information+Processing+Systems%2C+36%2E+ArXiv%3A2202%2E05262%2E)
25. Kevin Meng, Arnab Sen Sharma, Alex Andonian, Yonatan Belinkov, and David Bau. 2023. Mass editing memory in a transformer. The Eleventh International Conference on Learning Representations (ICLR). [find](https://scholar.google.com/scholar?q=Kevin+Meng%2C+Arnab+Sen+Sharma%2C+Alex+Andonian%2C+Yonatan+Belinkov%2C+and+David+Bau%2E+2023%2E+Mass+editing+memory+in+a+transformer%2E+The+Eleventh+International+Conference+on+Learning+Representations+%28ICLR%29%2E)
26. Junho Myung, Nayeon Lee, Yi Zhou, Jiho Jin, Rifki Afina Putri, Dimosthenis Antypas, Hsuvas Borkakoty, Eunsu Kim, Carla Perez-Almendros, Abinew Ali Ayele, Víctor Gutiérrez-Basulto, Yazmín Ibáñez García, Hwaran Lee, Shamsuddeen Hassan Muhammad, Kiwoong Park, Anar Sabuhi Rzayev, Nina White, Seid Muhie… — [BLEnD: A Benchmark for LLMs on Everyday Knowledge in Diverse Cultures and Languages](https://md.research.elamur.ai/free/en/papers/myung-2024-blend-benchmark-llms-everyday-knowledge/)
27. Junho Myung, Nayeon Lee, Yi Zhou, Jiho Jin, Rifki Afina Putri, Dimosthenis Antypas, Hsuvas Borkakoty, Eunsu Kim, Carla Perez-Almendros, Abinew Ali Ayele, Víctor Gutiérrez-Basulto, Yazmín Ibáñez-García, Hwaran Lee, Shamsuddeen Hassan Muhammad, Kiwoong Park, Anar Sabuhi Rzayev, Nina White, Seid Muhie… — [BLEnD: A Benchmark for LLMs on Everyday Knowledge in Diverse Cultures and Languages](https://md.research.elamur.ai/free/en/papers/myung-2024-blend-benchmark-llms-everyday-knowledge/)
28. Moin Nadeem, Anna Bethke, and Siva Reddy. 2021. [StereoSet: Measuring stereotypical bias in pretrained language models](https://md.research.elamur.ai/free/en/papers/nadeem-2021-stereoset-measuring-stereotypical-bias-pretrained/). In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long…
29. Nikita Nangia, Clara Vania, Rasika Bhalerao, and Samuel R. Bowman. 2020. [CrowS-pairs: A challenge dataset for measuring social biases in masked language models](https://md.research.elamur.ai/free/en/papers/nangia-2020-crows-pairs-challenge-dataset-measuring/). In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1953–1967, Online. Association for…
30. Christopher Olah, Nick Cammarata, Ludwig Schubert, Gabriel Goh, Michael Petrov, and Shan Carter. 2020. Zoom in: An introduction to circuits. [find](https://scholar.google.com/scholar?q=Christopher+Olah%2C+Nick+Cammarata%2C+Ludwig+Schubert%2C+Gabriel+Goh%2C+Michael+Petrov%2C+and+Shan+Carter%2E+2020%2E+Zoom+in%3A+An+introduction+to+circuits%2E)
31. Pew Global Attitudes Survey. 2022. Pew global attitudes survey. [https://www.pewresearch.org/](https://www.pewresearch.org/). [↗](https://www.pewresearch.org/)
32. Abhinav Sukumar Rao, Akhila Yerukola, Vishwa Shah, Katharina Reinecke, and Maarten Sap. 2025. [NormAd: A framework for measuring the cultural adaptability of large language models](https://md.research.elamur.ai/free/en/papers/rao-2025-normad-framework-measuring-cultural-adaptability/). In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Lingui…
33. Timo Schick, Sahana Udupa, and Hinrich Schütze. 2021. [Self-diagnosis and self-debiasing: A proposal for reducing corpus-based bias in nlp](https://md.research.elamur.ai/free/en/papers/schick-2021-self-diagnosis-self-debiasing-proposal/). Transactions of the Association for Computational Linguistics, 9:1408– 1424.
34. Wonduk Seo, Zonghao Yuan, and Yi Bu. 2025. Valuesrag: Enhancing cultural alignment through retrieval-augmented contextual learning. Preprint, arXiv:2501.01031. — [arXiv:2501.01031](https://arxiv.org/abs/2501.01031) [find](https://scholar.google.com/scholar?q=Wonduk+Seo%2C+Zonghao+Yuan%2C+and+Yi+Bu%2E+2025%2E+Valuesrag%3A+Enhancing+cultural+alignment+through+retrieval%2Daugmented+contextual+learning%2E+Preprint%2C+arXiv%3A2501%2E01031%2E+%E2%80%94+%5BarXiv%3A2501%2E01031%5D%28https%3A%2F%2Farxiv%2Eorg%2Fa)
35. Weiyan Shi, Ryan Li, Yutong Zhang, Caleb Ziems, Sunny Yu, Raya Horesh, Rogério Abreu De Paula, and Diyi Yang. 2024. [CultureBank: An online community-driven knowledge base towards culturally aware language technologies](https://md.research.elamur.ai/free/en/papers/shi-2024-culturebank-online-community-driven-knowledge/). In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 4…
36. Shivalika Singh, Angelika Romanou, Clémentine Fourrier, David Ifeoluwa Adelani, Jian Gang Ngui, Daniel Vila-Suero, Peerat Limkonchotiwat, Kelly Marchisio, Wei Qi Leong, Yosephine Susanto, Raymond Ng, Shayne Longpre, Sebastian Ruder, Wei-Yin Ko, Antoine Bosselut, Alice Oh, Andre Martins, Leshem Chosh… — [Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation](https://md.research.elamur.ai/free/en/papers/singh-2025-global-mmlu-understanding-addressing-cultural/)
37. Eric Michael Smith, Melissa Hall, Melanie Kambadur, Eleonora Presani, and Adina Williams. 2022. [“I’m sorry to hear that”: Finding new biases in language models with a holistic descriptor dataset](https://md.research.elamur.ai/free/en/papers/smith-2022-i-m-sorry-hear-that/). In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 9180–92…
38. Yi Su, Jiayi Zhang, Shu Yang, Xinhai Wang, Lijie Hu, and Di Wang. 2025. Understanding how value neurons shape the generation of specified values in LLMs. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 9433–9452, Suzhou, China. Association for Computational Linguistic… [find](https://scholar.google.com/scholar?q=Yi+Su%2C+Jiayi+Zhang%2C+Shu+Yang%2C+Xinhai+Wang%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025%2E+Understanding+how+value+neurons+shape+the+generation+of+specified+values+in+LLMs%2E+In+Findings+of+the+Association+for+Computation)
39. Wenjie Sun, Di Wang, and Lijie Hu. 2026. The price of amortized inference in sparse autoencoders. In The Fourteenth International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Wenjie+Sun%2C+Di+Wang%2C+and+Lijie+Hu%2E+2026%2E+The+price+of+amortized+inference+in+sparse+autoencoders%2E+In+The+Fourteenth+International+Conference+on+Learning+Representations%2E)
40. Yan Tao, Olga Viberg, Ryan S Baker, and René F Kizilcec. 2024. Cultural bias and cultural alignment of large language models. PNAS Nexus, 3(9):pgae346. [find](https://scholar.google.com/scholar?q=Yan+Tao%2C+Olga+Viberg%2C+Ryan+S+Baker%2C+and+Ren%C3%A9+F+Kizilcec%2E+2024%2E+Cultural+bias+and+cultural+alignment+of+large+language+models%2E+PNAS+Nexus%2C+3%289%29%3Apgae346%2E)
41. Adly Templeton, Tom Conerly, Jonathan Marcus, Jack Lindsey, Trenton Bricken, Brian Chen, Adam Pearce, Craig Citro, Emmanuel Ameisen, Andy Jones, Hoagy Cunningham, Nicholas L Turner, Callum McDougall, Monte MacDiarmid, C. Daniel Freeman, Theodore R. Sumers, Edward Rees, Joshua Batson, Adam Jermyn, Sh… [find](https://scholar.google.com/scholar?q=Adly+Templeton%2C+Tom+Conerly%2C+Jonathan+Marcus%2C+Jack+Lindsey%2C+Trenton+Bricken%2C+Brian+Chen%2C+Adam+Pearce%2C+Craig+Citro%2C+Emmanuel+Ameisen%2C+Andy+Jones%2C+Hoagy+Cunningham%2C+Nicholas+L+Turner%2C+Callum+McDougall%2C+)
42. Alexander Matt Turner, Lisa Thiergart, Gavin Leech, David Udell, Juan J. Vazquez, Ulisse Mini, and Monte MacDiarmid. 2024. [Steering language models with activation engineering](https://md.research.elamur.ai/free/en/papers/turner-2023-steering-language-models-activation-engineering/). Preprint, arXiv:2308.10248.
43. Kevin Wang, Alexandre Variengien, Arthur Conmy, Buck Shlegeris, and Jacob Steinhardt. 2022. Interpretability in the wild: a circuit for indirect object identification in gpt-2 small. Preprint, arXiv:2211.00593. [find](https://scholar.google.com/scholar?q=Kevin+Wang%2C+Alexandre+Variengien%2C+Arthur+Conmy%2C+Buck+Shlegeris%2C+and+Jacob+Steinhardt%2E+2022%2E+Interpretability+in+the+wild%3A+a+circuit+for+indirect+object+identification+in+gpt%2D2+small%2E+Preprint%2C+arXiv%3A2)
44. Wenxuan Wang, Wenxiang Jiao, Jingyuan Huang, Ruyi Dai, Jen-tse Huang, Zhaopeng Tu, and Michael Lyu. 2024. [Not all countries celebrate thanksgiving: On the cultural dominance in large language models](https://md.research.elamur.ai/free/en/papers/wang-2024-not-all-countries-celebrate-thanksgiving/). In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1…
45. Xinhai Wang, Shu Yang, Liangyu Wang, Lin Zhang, Huanyi Xie, Lijie Hu, and Di Wang. 2025. Pahq: Accelerating automated circuit discovery through mixed-precision inference optimization. arXiv preprint arXiv:2510.23264. — [arXiv:2510.23264](https://arxiv.org/abs/2510.23264) [find](https://scholar.google.com/scholar?q=Xinhai+Wang%2C+Shu+Yang%2C+Liangyu+Wang%2C+Lin+Zhang%2C+Huanyi+Xie%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025%2E+Pahq%3A+Accelerating+automated+circuit+discovery+through+mixed%2Dprecision+inference+optimization%2E+arXiv+preprint+a)
46. World Values Survey. 2022. World values survey. [https://www.worldvaluessurvey.org/wvs.jsp](https://www.worldvaluessurvey.org/wvs.jsp). [↗](https://www.worldvaluessurvey.org/wvs.jsp)
47. Xuansheng Wu, Haiyan Zhao, Yaochen Zhu, Yucheng Shi, Fan Yang, Lijie Hu, Tianming Liu, Xiaoming Zhai, Wenlin Yao, Jundong Li, et al. 2024. Usable xai: 10 strategies towards exploiting explainability in the llm era. arXiv preprint arXiv:2403.08946. — [arXiv:2403.08946](https://arxiv.org/abs/2403.08946) [find](https://scholar.google.com/scholar?q=Xuansheng+Wu%2C+Haiyan+Zhao%2C+Yaochen+Zhu%2C+Yucheng+Shi%2C+Fan+Yang%2C+Lijie+Hu%2C+Tianming+Liu%2C+Xiaoming+Zhai%2C+Wenlin+Yao%2C+Jundong+Li%2C+et+al%2E+2024%2E+Usable+xai%3A+10+strategies+towards+exploiting+explainability+i)
48. Taisei Yamamoto, Ryoma Kumon, Danushka Bollegala, and Hitomi Yanaka. 2025. Neuron-level analysis of cultural understanding in large language models. Preprint, arXiv:2510.08284. — [arXiv:2510.08284](https://arxiv.org/abs/2510.08284) [find](https://scholar.google.com/scholar?q=Taisei+Yamamoto%2C+Ryoma+Kumon%2C+Danushka+Bollegala%2C+and+Hitomi+Yanaka%2E+2025%2E+Neuron%2Dlevel+analysis+of+cultural+understanding+in+large+language+models%2E+Preprint%2C+arXiv%3A2510%2E08284%2E+%E2%80%94+%5BarXiv%3A2510%2E08284%5D%28ht)
49. Shu Yang, Shenzhe Zhu, Zeyu Wu, Keyu Wang, Junchi Yao, Junchao Wu, Lijie Hu, Mengdi Li, Derek F Wong, and Di Wang. 2025. Fraud-r1: A multi-round benchmark for assessing the robustness of llm against augmented fraud and phishing inducements. In Findings of the Association for Computational Linguistic… [find](https://scholar.google.com/scholar?q=Shu+Yang%2C+Shenzhe+Zhu%2C+Zeyu+Wu%2C+Keyu+Wang%2C+Junchi+Yao%2C+Junchao+Wu%2C+Lijie+Hu%2C+Mengdi+Li%2C+Derek+F+Wong%2C+and+Di+Wang%2E+2025%2E+Fraud%2Dr1%3A+A+multi%2Dround+benchmark+for+assessing+the+robustness+of+llm+against+a)
50. Junchi Yao, Shu Yang, Jianhua Xu, Lijie Hu, Mengdi Li, and Di Wang. 2025. Understanding the repeat curse in large language models from a feature perspective. In Findings of the Association for Computational Linguistics: ACL 2025, pages 7787–7815. [find](https://scholar.google.com/scholar?q=Junchi+Yao%2C+Shu+Yang%2C+Jianhua+Xu%2C+Lijie+Hu%2C+Mengdi+Li%2C+and+Di+Wang%2E+2025%2E+Understanding+the+repeat+curse+in+large+language+models+from+a+feature+perspective%2E+In+Findings+of+the+Association+for+Computa)
51. Da Yin, Hritik Bansal, Masoud Monajatipoor, Liunian Harold Li, and Kai-Wei Chang. 2022. Geomlama: Geo-diverse commonsense probing on multilingual pre-trained language models. In EMNLP. [find](https://scholar.google.com/scholar?q=Da+Yin%2C+Hritik+Bansal%2C+Masoud+Monajatipoor%2C+Liunian+Harold+Li%2C+and+Kai%2DWei+Chang%2E+2022%2E+Geomlama%3A+Geo%2Ddiverse+commonsense+probing+on+multilingual+pre%2Dtrained+language+models%2E+In+EMNLP%2E)
52. Haeun Yu, Seogyeong Jeong, Siddhesh Pawar, Jisu Shin, Jiho Jin, Junho Myung, Alice Oh, and Isabelle Augenstein. 2025a. Entangled in representations: Mechanistic investigation of cultural biases in large language models. Preprint, arXiv:2508.08879. [find](https://scholar.google.com/scholar?q=Haeun+Yu%2C+Seogyeong+Jeong%2C+Siddhesh+Pawar%2C+Jisu+Shin%2C+Jiho+Jin%2C+Junho+Myung%2C+Alice+Oh%2C+and+Isabelle+Augenstein%2E+2025a%2E+Entangled+in+representations%3A+Mechanistic+investigation+of+cultural+biases+in+lar)
53. Manjiang Yu, Hongji Li, Priyanka Singh, Xue Li, Di Wang, and Lijie Hu. 2025b. Pixel: Adaptive steering via position-wise injection with exact estimated levels under subspace calibration. arXiv preprint arXiv:2510.10205. — [arXiv:2510.10205](https://arxiv.org/abs/2510.10205) [find](https://scholar.google.com/scholar?q=Manjiang+Yu%2C+Hongji+Li%2C+Priyanka+Singh%2C+Xue+Li%2C+Di+Wang%2C+and+Lijie+Hu%2E+2025b%2E+Pixel%3A+Adaptive+steering+via+position%2Dwise+injection+with+exact+estimated+levels+under+subspace+calibration%2E+arXiv+preprin)
54. Fred Zhang and Neel Nanda. 2023. Towards best practices of activation patching in language models: Metrics and methods. ICLR. [find](https://scholar.google.com/scholar?q=Fred+Zhang+and+Neel+Nanda%2E+2023%2E+Towards+best+practices+of+activation+patching+in+language+models%3A+Metrics+and+methods%2E+ICLR%2E)
55. Hongbin Zhang, Kehai Chen, Xuefeng Bai, Yang Xiang, and Min Zhang. 2025a. Evaluating and improving cultural awareness of reward models for llm alignment. Preprint, arXiv:2509.21798. — [arXiv:2509.21798](https://arxiv.org/abs/2509.21798) [find](https://scholar.google.com/scholar?q=Hongbin+Zhang%2C+Kehai+Chen%2C+Xuefeng+Bai%2C+Yang+Xiang%2C+and+Min+Zhang%2E+2025a%2E+Evaluating+and+improving+cultural+awareness+of+reward+models+for+llm+alignment%2E+Preprint%2C+arXiv%3A2509%2E21798%2E+%E2%80%94+%5BarXiv%3A2509%2E2179)
56. Lin Zhang, Wenshuo Dong, Zhuoran Zhang, Shu Yang, Lijie Hu, Ninghao Liu, Pan Zhou, and Di Wang. Eap-gp: Mitigating saturation effect in gradient-based automated circuit identification. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Lin+Zhang%2C+Wenshuo+Dong%2C+Zhuoran+Zhang%2C+Shu+Yang%2C+Lijie+Hu%2C+Ninghao+Liu%2C+Pan+Zhou%2C+and+Di+Wang%2E+Eap%2Dgp%3A+Mitigating+saturation+effect+in+gradient%2Dbased+automated+circuit+identification%2E+In+The+Thirty%2Dn)
57. Lin Zhang, Lijie Hu, and Di Wang. 2025b. Mechanistic unveiling of transformer circuits: Self-influence as a key to model reasoning. In Findings of the Association for Computational Linguistics: NAACL 2025, pages 1387–1404. [find](https://scholar.google.com/scholar?q=Lin+Zhang%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025b%2E+Mechanistic+unveiling+of+transformer+circuits%3A+Self%2Dinfluence+as+a+key+to+model+reasoning%2E+In+Findings+of+the+Association+for+Computational+Linguistics%3A+NAACL+)
58. Zhuoran Zhang, Yongxiang Li, Zijian Kan, Keyuan Cheng, Lijie Hu, and Di Wang. 2025c. Locate-then-edit for multi-hop factual recall under knowledge editing. In International Conference on Machine Learning, pages 75369–75391. PMLR. across the 5 rewrites. - The rewrites do not need to remain questions;… [find](https://scholar.google.com/scholar?q=Zhuoran+Zhang%2C+Yongxiang+Li%2C+Zijian+Kan%2C+Keyuan+Cheng%2C+Lijie+Hu%2C+and+Di+Wang%2E+2025c%2E+Locate%2Dthen%2Dedit+for+multi%2Dhop+factual+recall+under+knowledge+editing%2E+In+International+Conference+on+Machine+Learn)
59. Output only the 5 rewritten sentences, each on a separate line. B [find](https://scholar.google.com/scholar?q=Output+only+the+5+rewritten+sentences%2C+each+on+a+separate+line%2E+B)

*Updated 31 October 2026*