---
title: "Revitalizing Black-Box Interpretability — paper by Liu et al. (2026)"
description: "The authors build on the observation that different LLMs tend to behave alike on similar inputs. 53 references."
authors: ["Junhao Liu", "Haonan Yu", "Zhenyu Yan", "Xin Zhang"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.220/"
doi: "10.18653/v1/2026.acl-long.220"
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/liu-2026-revitalizing-black-box-interpretability-actionable/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/liu-2026-revitalizing-black-box-interpretability-actionable/)

[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/48/) / Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models

# Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models

**Authors:** [Junhao Liu](https://md.research.elamur.ai/free/en/authors/junhao-liu/), Haonan Yu, Zhenyu Yan and [Xin Zhang](https://md.research.elamur.ai/free/en/authors/xin-zhang/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/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.acl-long.220/) · [PDF](https://aclanthology.org/2026.acl-long.220.pdf) · [DOI 10.18653/v1/2026.acl-long.220](https://doi.org/10.18653/v1/2026.acl-long.220) · [OpenAlex](https://openalex.org/W4417301055)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Perturbation-based explainers such as LIME and SHAP need thousands of model queries for each explanation. That makes them too expensive for closed API models like GPT-4o, especially when explanations are meant to guide optimization rather than just inspection. The authors build on the observation that different LLMs tend to behave alike on similar inputs. They compute explanations on a cheap local model, such as Qwen or LLaMA, and transfer them to the expensive target.

To guard against cases where the two models diverge, they add a screen-and-apply procedure with two checks:
- an offline, task-level sequential one-sided paired t-test, which confirms that proxy fidelity reaches at least a fraction τ of oracle fidelity;
- an online, per-instance check that both models give the same prediction.

The evaluation covers 12 LLMs and seven tasks. Proxy explanations keep over 90% of oracle fidelity at roughly 9.5% of the cost. They also work in practice for pruning in-context examples and for filtering out poisoned ones. This turns black-box interpretability into an affordable tool for LLM development.

## Key points
- **Cost motivation:** explaining 50 validation inputs with 1,000 perturbations each means 50,000 GPT-4o calls, costing more than $100.
- **Fidelity and cost:** proxy explanations reach over 90% fidelity relative to oracle explanations while cutting cost by 90.5%. The evaluation spans models from LLaMA 3.1 and DeepSeek V3 to GPT-4o, and tasks including MMLU.
- **Prompt compression:** KSHAP-based removal of in-context examples was tested for GPT-4o on MMLU subjects, HellaSwag, GSM8K and PIQA, keeping at least 90% of full-prompt performance. Qwen 2.5 proxies achieved on average 91.3% of the oracle's compression ratio. They outperformed random deletion, AttnComp and LLMLingua.
- **Poisoned example removal:** GPT-4o accuracy with proxy-guided removal versus oracle-guided removal:

  | Dataset | Proxy | Oracle | Random deletion |
  |---|---|---|---|
  | SST-2 | 94.0% | 94.2% | 87.1% |
  | HellaSwag | 93.5% | 93.7% | 88.4% |
  | PIQA | 90.7% | 91.5% | 79.6% |

- **Released resource:** XLLM-Bench, a dataset of perturbation samples and model outputs, available with the code at github.com/outerform/XLLM-Bench.

*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

- [Peking University](https://md.research.elamur.ai/free/en/universities/peking-university/) — University, Beijing, China · 315 works · 43,268 citations

📚 Shelves (3): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [💻 Software](https://md.research.elamur.ai/free/en/shelves/computer-science/software/) · 119 works

🧭 Topics (2): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,894 works · [🔬 Interpretability](https://md.research.elamur.ai/free/en/topics/interpretability/) · 149 works

## Cite

Cite as: Junhao Liu, Haonan Yu, Zhenyu Yan and Xin Zhang. 2026. Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

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

```bibtex
@inproceedings{liu2026revitalizing,
  title = {{Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models}},
  author = {Junhao Liu, Haonan Yu, Zhenyu Yan and Xin Zhang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.220},
  url = {https://doi.org/10.18653/v1/2026.acl-long.220}
}
```

- [← previous Revisiting the Systematicity in Negation in the Era of In-Context Learning](https://md.research.elamur.ai/free/en/papers/yanaka-2026-revisiting-systematicity-negation-era-context/)
- [next → Reviving Iterative Refinement in Diffusion-based NER with an Initializer-Restorer Approach](https://md.research.elamur.ai/free/en/papers/trieu-2026-reviving-iterative-refinement-diffusion-based/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/48/)

## References

53 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Aparna Balagopalan, Haoran Zhang, Kimia Hamidieh, Thomas Hartvigsen, Frank Rudzicz, and Marzyeh Ghassemi. 2022. The road to explainability is paved with bias: Measuring the fairness of explanations. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, pages 1194–… [find](https://scholar.google.com/scholar?q=Aparna+Balagopalan%2C+Haoran+Zhang%2C+Kimia+Hamidieh%2C+Thomas+Hartvigsen%2C+Frank+Rudzicz%2C+and+Marzyeh+Ghassemi%2E+2022%2E+The+road+to+explainability+is+paved+with+bias%3A+Measuring+the+fairness+of+explanations%2E+I)
2. Jonathan Berant, Andrew Chou, Roy Frostig, and Percy Liang. 2013. [Semantic parsing on Freebase from question-answer pairs](https://md.research.elamur.ai/free/en/papers/berant-2013-semantic-parsing-freebase-question-answer/). In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 1533–1544, Seattle, Washington, USA. Association for Computational Linguistics.
3. Yonatan Bisk, Rowan Zellers, Ronan Le Bras, Jianfeng Gao, and Yejin Choi. 2019. [Piqa: Reasoning about physical commonsense in natural language](https://md.research.elamur.ai/free/en/papers/bisk-2019-piqa-reasoning-about-physical-commonsense/). Preprint, arXiv:1911.11641.
4. Jianbo Chen, Le Song, Martin Wainwright, and Michael Jordan. 2018. Learning to explain: An information-theoretic perspective on model interpretation. In Proceedings of the 35th International Conference on Machine Learning, volume 80 of Proceedings of Machine Learning Research, pages 883–892. PMLR. [find](https://scholar.google.com/scholar?q=Jianbo+Chen%2C+Le+Song%2C+Martin+Wainwright%2C+and+Michael+Jordan%2E+2018%2E+Learning+to+explain%3A+An+information%2Dtheoretic+perspective+on+model+interpretation%2E+In+Proceedings+of+the+35th+International+Conferenc)
5. Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. [Training verifiers to solve math word problems](https://md.research.elamur.ai/free/en/papers/cobbe-2021-training-verifiers-solve-math-word/). Preprint, arXiv:2110.14168.
6. Ian Covert, Chanwoo Kim, Su-In Lee, James Zou, and Tatsunori Hashimoto. 2024. Stochastic amortization: A unified approach to accelerate feature and data attribution. In Advances in Neural Information Processing Systems, volume 37, page 4374–4423. Curran Associates, Inc. [find](https://scholar.google.com/scholar?q=Ian+Covert%2C+Chanwoo+Kim%2C+Su%2DIn+Lee%2C+James+Zou%2C+and+Tatsunori+Hashimoto%2E+2024%2E+Stochastic+amortization%3A+A+unified+approach+to+accelerate+feature+and+data+attribution%2E+In+Advances+in+Neural+Information+)
7. David Dandolo, Chiara Masiero, Mattia Carletti, Davide Dalle Pezze, and Gian Antonio Susto. 2023. Acme—accelerated model-agnostic explanations: Fast whitening of the machine-learning black box. Expert Systems with Applications, 214:119115. [find](https://scholar.google.com/scholar?q=David+Dandolo%2C+Chiara+Masiero%2C+Mattia+Carletti%2C+Davide+Dalle+Pezze%2C+and+Gian+Antonio+Susto%2E+2023%2E+Acme%E2%80%94accelerated+model%2Dagnostic+explanations%3A+Fast+whitening+of+the+machine%2Dlearning+black+box%2E+Expert)
8. DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, and 181 others. 2025. Deepseek-v3 technical report. Preprint, arXiv:2… [find](https://scholar.google.com/scholar?q=DeepSeek%2DAI%2C+Aixin+Liu%2C+Bei+Feng%2C+Bing+Xue%2C+Bingxuan+Wang%2C+Bochao+Wu%2C+Chengda+Lu%2C+Chenggang+Zhao%2C+Chengqi+Deng%2C+Chenyu+Zhang%2C+Chong+Ruan%2C+Damai+Dai%2C+Daya+Guo%2C+Dejian+Yang%2C+Deli+Chen%2C+Dongjie+Ji%2C+Erhan)
9. Rudresh Dwivedi, Devam Dave, Het Naik, Smiti Singhal, Rana Omer, Pankesh Patel, Bin Qian, Zhenyu Wen, Tejal Shah, and Graham Morgan. 2023. Explainable ai (xai): Core ideas, techniques, and solutions. ACM Computing Surveys, 55(9):1–33. [find](https://scholar.google.com/scholar?q=Rudresh+Dwivedi%2C+Devam+Dave%2C+Het+Naik%2C+Smiti+Singhal%2C+Rana+Omer%2C+Pankesh+Patel%2C+Bin+Qian%2C+Zhenyu+Wen%2C+Tejal+Shah%2C+and+Graham+Morgan%2E+2023%2E+Explainable+ai+%28xai%29%3A+Core+ideas%2C+techniques%2C+and+solutions%2E+)
10. James Enouen, Hootan Nakhost, Sayna Ebrahimi, Sercan Arik, Yan Liu, and Tomas Pfister. 2024. TextGen-SHAP: Scalable post-hoc explanations in text generation with long documents. In Findings of the Association for Computational Linguistics: ACL 2024, pages 13984–14011, Bangkok, Thailand. Association… [find](https://scholar.google.com/scholar?q=James+Enouen%2C+Hootan+Nakhost%2C+Sayna+Ebrahimi%2C+Sercan+Arik%2C+Yan+Liu%2C+and+Tomas+Pfister%2E+2024%2E+TextGen%2DSHAP%3A+Scalable+post%2Dhoc+explanations+in+text+generation+with+long+documents%2E+In+Findings+of+the+Ass)
11. Gemini Team Google, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M. Dai, Anja Hauth, Katie Millican, David Silver, Melvin Johnson, Ioannis Antonoglou, Julian Schrittwieser, Amelia Glaese, Jilin Chen, Emily Pitler, Timothy Lillicrap, Angeliki… — [Gemini: A Family of Highly Capable Multimodal Models](https://md.research.elamur.ai/free/en/papers/gemini-team-2023-gemini-family-highly-capable-multimodal/)
12. Yu-Neng Chuang, Guanchu Wang, Fan Yang, Zirui Liu, Xuanting Cai, Mengnan Du, and Xia Hu. 2023a. Efficient xai techniques: A taxonomic survey. Preprint, arXiv:2302.03225. — [arXiv:2302.03225](https://arxiv.org/abs/2302.03225) [find](https://scholar.google.com/scholar?q=Yu%2DNeng+Chuang%2C+Guanchu+Wang%2C+Fan+Yang%2C+Zirui+Liu%2C+Xuanting+Cai%2C+Mengnan+Du%2C+and+Xia+Hu%2E+2023a%2E+Efficient+xai+techniques%3A+A+taxonomic+survey%2E+Preprint%2C+arXiv%3A2302%2E03225%2E+%E2%80%94+%5BarXiv%3A2302%2E03225%5D%28https%3A%2F%2Fa)
13. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, and 542 o… — [The Llama 3 Herd of Models](https://md.research.elamur.ai/free/en/papers/grattafiori-2024-llama-3-herd-models/)
14. Yu-Neng Chuang, Guanchu Wang, Fan Yang, Quan Zhou, Pushkar Tripathi, Xuanting Cai, and Xia Hu. 2023b. Cortx: Contrastive framework for real-time explanation. Preprint, arXiv:2303.02794. [find](https://scholar.google.com/scholar?q=Yu%2DNeng+Chuang%2C+Guanchu+Wang%2C+Fan+Yang%2C+Quan+Zhou%2C+Pushkar+Tripathi%2C+Xuanting+Cai%2C+and+Xia+Hu%2E+2023b%2E+Cortx%3A+Contrastive+framework+for+real%2Dtime+explanation%2E+Preprint%2C+arXiv%3A2303%2E02794%2E)
15. Stefan Hackmann, Haniyeh Mahmoudian, Mark Stead-man, and Michael Schmidt. 2024. Word importance explains how prompts affect language model outputs. Preprint, arXiv:2403.03028. — [arXiv:2403.03028](https://arxiv.org/abs/2403.03028) [find](https://scholar.google.com/scholar?q=Stefan+Hackmann%2C+Haniyeh+Mahmoudian%2C+Mark+Stead%2Dman%2C+and+Michael+Schmidt%2E+2024%2E+Word+importance+explains+how+prompts+affect+language+model+outputs%2E+Preprint%2C+arXiv%3A2403%2E03028%2E+%E2%80%94+%5BarXiv%3A2403%2E03028%5D%28htt)
16. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. [Measuring massive multitask language understanding](https://md.research.elamur.ai/free/en/papers/hendrycks-2020-measuring-massive-multitask-language-understanding/). arXiv preprint arXiv:2009.03300.
17. Aya Abdelsalam Ismail, Hector Corrada Bravo, and Soheil Feizi. 2021. Improving deep learning interpretability by saliency guided training. Advances in Neural Information Processing Systems, 34:26726– 26739. [find](https://scholar.google.com/scholar?q=Aya+Abdelsalam+Ismail%2C+Hector+Corrada+Bravo%2C+and+Soheil+Feizi%2E+2021%2E+Improving+deep+learning+interpretability+by+saliency+guided+training%2E+Advances+in+Neural+Information+Processing+Systems%2C+34%3A26726%E2%80%93+)
18. Neil Jethani, Mukund Sudarshan, Ian Connick Covert, Su-In Lee, and Rajesh Ranganath. 2022. FastSHAP: Real-time shapley value estimation. In International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Neil+Jethani%2C+Mukund+Sudarshan%2C+Ian+Connick+Covert%2C+Su%2DIn+Lee%2C+and+Rajesh+Ranganath%2E+2022%2E+FastSHAP%3A+Real%2Dtime+shapley+value+estimation%2E+In+International+Conference+on+Learning+Representations%2E)
19. Huiqiang Jiang, Qianhui Wu, Chin-Yew Lin, Yuqing Yang, and Lili Qiu. 2023. [Llmlingua: Compressing prompts for accelerated inference of large language models](https://md.research.elamur.ai/free/en/papers/jiang-2023-llmlingua-compressing-prompts-accelerated-inference/). Preprint, arXiv:2310.05736.
20. Liwei Jiang, Yuanjun Chai, Margaret Li, Mickel Liu, Raymond Fok, Nouha Dziri, Yulia Tsvetkov, Maarten Sap, Alon Albalak, and Yejin Choi. 2025. Artificial hivemind: The open-ended homogeneity of language models (and beyond). arXiv preprint arXiv:2510.22954. — [arXiv:2510.22954](https://arxiv.org/abs/2510.22954) [find](https://scholar.google.com/scholar?q=Liwei+Jiang%2C+Yuanjun+Chai%2C+Margaret+Li%2C+Mickel+Liu%2C+Raymond+Fok%2C+Nouha+Dziri%2C+Yulia+Tsvetkov%2C+Maarten+Sap%2C+Alon+Albalak%2C+and+Yejin+Choi%2E+2025%2E+Artificial+hivemind%3A+The+open%2Dended+homogeneity+of+langua)
21. Martin Jullum, Annabelle Redelmeier, and Kjersti Aas. 2021. groupshapley: Efficient prediction explanation with shapley values for feature groups. arXiv preprint arXiv:2106.12228. — [arXiv:2106.12228](https://arxiv.org/abs/2106.12228) [find](https://scholar.google.com/scholar?q=Martin+Jullum%2C+Annabelle+Redelmeier%2C+and+Kjersti+Aas%2E+2021%2E+groupshapley%3A+Efficient+prediction+explanation+with+shapley+values+for+feature+groups%2E+arXiv+preprint+arXiv%3A2106%2E12228%2E+%E2%80%94+%5BarXiv%3A2106%2E12228%5D)
22. Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc Le, and Slav Petrov. 2019. [Natural…](https://md.research.elamur.ai/free/en/papers/kwiatkowski-2019-natural-questions-benchmark-question-answering/)
23. Sunbowen Lee, Junting Zhou, Chang Ao, Kaige Li, Xinrun Du, Sirui He, Haihong Wu, Tianci Liu, Jiaheng Liu, Hamid Alinejad-Rokny, Min Yang, Yitao Liang, Zhoufutu Wen, and Shiwen Ni. 2025. Quantification of large language model distillation. (arXiv:2501.12619). ArXiv:2501.12619 [cs]. [find](https://scholar.google.com/scholar?q=Sunbowen+Lee%2C+Junting+Zhou%2C+Chang+Ao%2C+Kaige+Li%2C+Xinrun+Du%2C+Sirui+He%2C+Haihong+Wu%2C+Tianci+Liu%2C+Jiaheng+Liu%2C+Hamid+Alinejad%2DRokny%2C+Min+Yang%2C+Yitao+Liang%2C+Zhoufutu+Wen%2C+and+Shiwen+Ni%2E+2025%2E+Quantification)
24. Junhao Liu, Haonan Yu, Zhenyu Yan, and Xin Zhang. 2026a. Focus-lime: Surgical interpretation of long-context large language models via proxy-based neighborhood selection. Preprint, arXiv:2602.04607. — [arXiv:2602.04607](https://arxiv.org/abs/2602.04607) [find](https://scholar.google.com/scholar?q=Junhao+Liu%2C+Haonan+Yu%2C+Zhenyu+Yan%2C+and+Xin+Zhang%2E+2026a%2E+Focus%2Dlime%3A+Surgical+interpretation+of+long%2Dcontext+large+language+models+via+proxy%2Dbased+neighborhood+selection%2E+Preprint%2C+arXiv%3A2602%2E04607%2E+%E2%80%94)
25. Junhao Liu, Haonan Yu, and Xin Zhang. 2026b. Beyond attribution: Unified concept-level explanations. Preprint, arXiv:2410.12439. — [arXiv:2410.12439](https://arxiv.org/abs/2410.12439) [find](https://scholar.google.com/scholar?q=Junhao+Liu%2C+Haonan+Yu%2C+and+Xin+Zhang%2E+2026b%2E+Beyond+attribution%3A+Unified+concept%2Dlevel+explanations%2E+Preprint%2C+arXiv%3A2410%2E12439%2E+%E2%80%94+%5BarXiv%3A2410%2E12439%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2410%2E12439%29)
26. Junhao Liu and Xin Zhang. 2025. Rex: A framework for incorporating temporal information in model-agnostic local explanation techniques. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 18888–18896. [find](https://scholar.google.com/scholar?q=Junhao+Liu+and+Xin+Zhang%2E+2025%2E+Rex%3A+A+framework+for+incorporating+temporal+information+in+model%2Dagnostic+local+explanation+techniques%2E+In+Proceedings+of+the+AAAI+Conference+on+Artificial+Intelligence)
27. Scott M. Lundberg and Su-In Lee. 2017. [A unified approach to interpreting model predictions](https://md.research.elamur.ai/free/en/papers/lundberg-2017-unified-approach-interpreting-model-predictions/). In Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems 2017, December 4-9, 2017, Long Beach, CA, USA, pages 4765–4774.
28. Ronny Luss, Erik Miehling, and Amit Dhurandhar. 2024. Cell your model: Contrastive explanations for large language models. arXiv preprint arXiv:2406.11785. — [arXiv:2406.11785](https://arxiv.org/abs/2406.11785) [find](https://scholar.google.com/scholar?q=Ronny+Luss%2C+Erik+Miehling%2C+and+Amit+Dhurandhar%2E+2024%2E+Cell+your+model%3A+Contrastive+explanations+for+large+language+models%2E+arXiv+preprint+arXiv%3A2406%2E11785%2E+%E2%80%94+%5BarXiv%3A2406%2E11785%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2)
29. Andrew L. Maas, Raymond E. Daly, Peter T. Pham, Dan Huang, Andrew Y. Ng, and Christopher Potts. 2011. [Learning word vectors for sentiment analysis](https://md.research.elamur.ai/free/en/papers/maas-2011-learning-word-vectors-sentiment-analysis/). In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 142–150, Portland, Oregon…
30. Manoj Kumar Mahto. 2024. Explainable artificial intelligence: Fundamentals, approaches, challenges, xai evaluation, and validation. In Explainable Artificial Intelligence for Autonomous Vehicles, pages 25–49. CRC Press. [find](https://scholar.google.com/scholar?q=Manoj+Kumar+Mahto%2E+2024%2E+Explainable+artificial+intelligence%3A+Fundamentals%2C+approaches%2C+challenges%2C+xai+evaluation%2C+and+validation%2E+In+Explainable+Artificial+Intelligence+for+Autonomous+Vehicles%2C+page)
31. Rory Mitchell, Joshua Cooper, Eibe Frank, and Geoffrey Holmes. 2022. Sampling permutations for shapley value estimation. Preprint, arXiv:2104.12199. — [arXiv:2104.12199](https://arxiv.org/abs/2104.12199) [find](https://scholar.google.com/scholar?q=Rory+Mitchell%2C+Joshua+Cooper%2C+Eibe+Frank%2C+and+Geoffrey+Holmes%2E+2022%2E+Sampling+permutations+for+shapley+value+estimation%2E+Preprint%2C+arXiv%3A2104%2E12199%2E+%E2%80%94+%5BarXiv%3A2104%2E12199%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2104%2E121)
32. OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mohammad Bavarian, Jeff Belgum, and 262 others.… — [GPT-4 Technical Report](https://md.research.elamur.ai/free/en/papers/openai-2023-gpt-4-technical-report/)
33. Lucas Monteiro Paes, Dennis Wei, Hyo Jin Do, Hendrik Strobelt, Ronny Luss, Amit Dhurandhar, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Werner Geyer, and Soumya Ghosh. 2024. Multi-level explanations for generative language models. Preprint, arXiv:2403.14459. [find](https://scholar.google.com/scholar?q=Lucas+Monteiro+Paes%2C+Dennis+Wei%2C+Hyo+Jin+Do%2C+Hendrik+Strobelt%2C+Ronny+Luss%2C+Amit+Dhurandhar%2C+Manish+Nagireddy%2C+Karthikeyan+Natesan+Ramamurthy%2C+Prasanna+Sattigeri%2C+Werner+Geyer%2C+and+Soumya+Ghosh%2E+2024%2E+)
34. Verónica Pérez-Rosas, Bennett Kleinberg, Alexandra Lefevre, and Rada Mihalcea. 2018. [Automatic detection of fake news](https://md.research.elamur.ai/free/en/papers/rosas-2018-automatic-detection-fake-news/). In Proceedings of the 27th International Conference on Computational Linguistics, pages 3391–3401, Santa Fe, New Mexico, USA. Association for Computational Linguistics.
35. Sudhanshu Ranjan, Chung-En Sun, Linbo Liu, and Tsui-Wei Weng. 2023. Fooling GPT with adversarial in-context examples for text classification. In R0-FoMo:Robustness of Few-shot and Zero-shot Learning in Large Foundation Models. [find](https://scholar.google.com/scholar?q=Sudhanshu+Ranjan%2C+Chung%2DEn+Sun%2C+Linbo+Liu%2C+and+Tsui%2DWei+Weng%2E+2023%2E+Fooling+GPT+with+adversarial+in%2Dcontext+examples+for+text+classification%2E+In+R0%2DFoMo%3ARobustness+of+Few%2Dshot+and+Zero%2Dshot+Learning+i)
36. Nils Reimers and Iryna Gurevych. 2019. [Sentence-bert: Sentence embeddings using siamese bert-networks](https://md.research.elamur.ai/free/en/papers/reimers-2019-sentence-bert-sentence-embeddings-siamese/). In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics.
37. Marco Túlio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. [“why should I trust you?": Explaining the predictions of any classifier](https://md.research.elamur.ai/free/en/papers/ribeiro-2016-why-should-i-trust-explaining/). In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, August 13-17, 2016, pages 1135– 1144.…
38. Thomas Rojat, Raphaël Puget, David Filliat, Javier Del Ser, Rodolphe Gelin, and Natalia Díaz-Rodríguez. 2021. Explainable artificial intelligence (xai) on timeseries data: A survey. arXiv preprint arXiv:2104.00950. — [arXiv:2104.00950](https://arxiv.org/abs/2104.00950) [find](https://scholar.google.com/scholar?q=Thomas+Rojat%2C+Rapha%C3%ABl+Puget%2C+David+Filliat%2C+Javier+Del+Ser%2C+Rodolphe+Gelin%2C+and+Natalia+D%C3%ADaz%2DRodr%C3%ADguez%2E+2021%2E+Explainable+artificial+intelligence+%28xai%29+on+timeseries+data%3A+A+survey%2E+arXiv+preprint+arX)
39. Haonan Yu, Junhao Liu, and Xin Zhang. 2026. Man-chors: Memorization-based acceleration of anchors via rule reuse and transformation. Preprint, arXiv:2502.11068. — [arXiv:2502.11068](https://arxiv.org/abs/2502.11068) [find](https://scholar.google.com/scholar?q=Haonan+Yu%2C+Junhao+Liu%2C+and+Xin+Zhang%2E+2026%2E+Man%2Dchors%3A+Memorization%2Dbased+acceleration+of+anchors+via+rule+reuse+and+transformation%2E+Preprint%2C+arXiv%3A2502%2E11068%2E+%E2%80%94+%5BarXiv%3A2502%2E11068%5D%28https%3A%2F%2Farxiv%2Eorg%2F)
40. Rowan Zellers, Ari Holtzman, Yonatan Bisk, Ali Farhadi, and Yejin Choi. 2019. Hellaswag: Can a machine really finish your sentence? Preprint, arXiv:1905.07830. [find](https://scholar.google.com/scholar?q=Rowan+Zellers%2C+Ari+Holtzman%2C+Yonatan+Bisk%2C+Ali+Farhadi%2C+and+Yejin+Choi%2E+2019%2E+Hellaswag%3A+Can+a+machine+really+finish+your+sentence%3F+Preprint%2C+arXiv%3A1905%2E07830%2E)
41. Taylor Shin, Yasaman Razeghi, Robert L. Logan IV, Eric Wallace, and Sameer Singh. 2020. [AutoPrompt: Eliciting knowledge from language models with automatically generated prompts](https://md.research.elamur.ai/free/en/papers/shin-2020-autoprompt-eliciting-knowledge-language-models/). In Empirical Methods in Natural Language Processing (EMNLP).
42. Yu Zhang, Peter Tiňo, Aleš Leonardis, and Ke Tang. 2021. A survey on neural network interpretability. IEEE Transactions on Emerging Topics in Computational Intelligence, 5(5):726–742. ArXiv:2012.14261 [cs]. — [arXiv:2012.14261](https://arxiv.org/abs/2012.14261) [find](https://scholar.google.com/scholar?q=Yu+Zhang%2C+Peter+Tin%CC%8Co%2C+Ale%C5%A1+Leonardis%2C+and+Ke+Tang%2E+2021%2E+A+survey+on+neural+network+interpretability%2E+IEEE+Transactions+on+Emerging+Topics+in+Computational+Intelligence%2C+5%285%29%3A726%E2%80%93742%2E+ArXiv%3A2012%2E1426)
43. Richard Socher, Alex Perelygin, Jean Wu, Jason Chuang, Christopher D Manning, Andrew Y Ng, and Christopher Potts. 2013. [Recursive deep models for semantic compositionality over a sentiment treebank](https://md.research.elamur.ai/free/en/papers/socher-2013-recursive-deep-models-semantic-compositionality/). In Proceedings of the 2013 conference on empirical methods in natural language processing, pages 1631…
44. Yunlong Zhao, Haoran Wu, and Bo Xu. 2025. Leveraging attention to effectively compress prompts for long-context llms. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 26048–26056. [find](https://scholar.google.com/scholar?q=Yunlong+Zhao%2C+Haoran+Wu%2C+and+Bo+Xu%2E+2025%2E+Leveraging+attention+to+effectively+compress+prompts+for+long%2Dcontext+llms%2E+In+Proceedings+of+the+AAAI+Conference+on+Artificial+Intelligence%2C+volume+39%2C+pages)
45. Ian Tenney, Ryan Mullins, Bin Du, Shree Pandya, Minsuk Kahng, and Lucas Dixon. 2024. Interactive prompt debugging with sequence salience. arXiv preprint arXiv:2404.07498. — [arXiv:2404.07498](https://arxiv.org/abs/2404.07498) [find](https://scholar.google.com/scholar?q=Ian+Tenney%2C+Ryan+Mullins%2C+Bin+Du%2C+Shree+Pandya%2C+Minsuk+Kahng%2C+and+Lucas+Dixon%2E+2024%2E+Interactive+prompt+debugging+with+sequence+salience%2E+arXiv+preprint+arXiv%3A2404%2E07498%2E+%E2%80%94+%5BarXiv%3A2404%2E07498%5D%28https%3A%2F%2F)
46. Yilun Zhou, Caiming Xiong, Silvio Savarese, and Chien-Sheng Wu. 2024. [Shared imagination: Llms hallucinate alike](https://md.research.elamur.ai/free/en/papers/zhou-2024-shared-imagination-llms-hallucinate-alike/). arXiv preprint arXiv:2407.16604.
47. Guanchu Wang, Yu-Neng Chuang, Mengnan Du, Fan Yang, Quan Zhou, Pushkar Tripathi, Xuanting Cai, and Xia Hu. 2022. Accelerating shapley explanation via contributive cooperator selection. In Proceedings of the 39th International Conference on Machine Learning, volume 162 of Proceedings of Machine Learn… [find](https://scholar.google.com/scholar?q=Guanchu+Wang%2C+Yu%2DNeng+Chuang%2C+Mengnan+Du%2C+Fan+Yang%2C+Quan+Zhou%2C+Pushkar+Tripathi%2C+Xuanting+Cai%2C+and+Xia+Hu%2E+2022%2E+Accelerating+shapley+explanation+via+contributive+cooperator+selection%2E+In+Proceedings+)
48. Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, and Ming Zhou. 2020. Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. Preprint, arXiv:2002.10957. [find](https://scholar.google.com/scholar?q=Wenhui+Wang%2C+Furu+Wei%2C+Li+Dong%2C+Hangbo+Bao%2C+Nan+Yang%2C+and+Ming+Zhou%2E+2020%2E+Minilm%3A+Deep+self%2Dattention+distillation+for+task%2Dagnostic+compression+of+pre%2Dtrained+transformers%2E+Preprint%2C+arXiv%3A2002%2E1095)
49. Dustin Wright, Sarah Masud, Jared Moore, Srishti Yadav, Maria Antoniak, Chan Young Park, and Isabelle Augenstein. 2025. Epistemic diversity and knowledge collapse in large language models. (arXiv:2510.04226). ArXiv:2510.04226 [cs]. — [arXiv:2510.04226](https://arxiv.org/abs/2510.04226) [find](https://scholar.google.com/scholar?q=Dustin+Wright%2C+Sarah+Masud%2C+Jared+Moore%2C+Srishti+Yadav%2C+Maria+Antoniak%2C+Chan+Young+Park%2C+and+Isabelle+Augenstein%2E+2025%2E+Epistemic+diversity+and+knowledge+collapse+in+large+language+models%2E+%28arXiv%3A2510)
50. Qwen: An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, and 23 others. 2025. Qwen2.5 technical report. Preprint, arX… [find](https://scholar.google.com/scholar?q=Qwen%3A+An+Yang%2C+Baosong+Yang%2C+Beichen+Zhang%2C+Binyuan+Hui%2C+Bo+Zheng%2C+Bowen+Yu%2C+Chengyuan+Li%2C+Dayiheng+Liu%2C+Fei+Huang%2C+Haoran+Wei%2C+Huan+Lin%2C+Jian+Yang%2C+Jianhong+Tu%2C+Jianwei+Zhang%2C+Jianxin+Yang%2C+Jiaxi+Yan)
51. Chih-Kuan Yeh, Cheng-Yu Hsieh, Arun Suggala, David I Inouye, and Pradeep K Ravikumar. 2019. On the (in) fidelity and sensitivity of explanations. Advances in Neural Information Processing Systems, 32. [find](https://scholar.google.com/scholar?q=Chih%2DKuan+Yeh%2C+Cheng%2DYu+Hsieh%2C+Arun+Suggala%2C+David+I+Inouye%2C+and+Pradeep+K+Ravikumar%2E+2019%2E+On+the+%28in%29+fidelity+and+sensitivity+of+explanations%2E+Advances+in+Neural+Information+Processing+Systems%2C+32%2E)
52. Jinsung Yoon, James Jordon, and Mihaela van der Schaar. 2019. INVASE: Instance-wise variable selection using neural networks. In International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Jinsung+Yoon%2C+James+Jordon%2C+and+Mihaela+van+der+Schaar%2E+2019%2E+INVASE%3A+Instance%2Dwise+variable+selection+using+neural+networks%2E+In+International+Conference+on+Learning+Representations%2E)
53. The Use of Large Language Models [find](https://scholar.google.com/scholar?q=The+Use+of+Large+Language+Models)

*Updated 1 November 2026*