Elamur Research⌕
☰
home / authors / Yangsibo Huang

Yangsibo Huang

Yangsibo Huang has 2 papers in our library, in 2023–2025, most of them on Language models & pretraining, Reasoning, Bias, fairness & safety; often with Bill Yuchen Lin, Bo Li, Chiyuan Zhang.

2 papers in our library · 2023–2025

OpenAlex · ORCID · Wikidata · h-index 10 · 50 works · 466 citations OpenAlex, 2026-10-14

📄 2 works in our library · 🔗 Cited 769 times in our library · 14 more in our library without a page yet · ⭐ Paper Stars ranking: #5563

The profile is the library's match of the name; two people with one name are not always told apart.

Elsewhere: Google Scholar · Semantic Scholar · dblp · ACL Anthology

 Affiliations

🎓Princeton UniversityPrinceton · United States2020–2025 · 24 papers🏢GoogleMountain View · United States2024–2025 · 5 papers🔬Institute for Advanced StudyPrinceton · United States2020 · 2 papers🎓University of California, Los AngelesLos Angeles · United States2019–2020 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 3 more institutions on single bylines are not listed.

 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Artificial Intelligence · 2 worksArts and Humanities · 2 worksComputer Science · 2 worksLanguage and Linguistics · 2 works

🧭 Topics 3Research subjects and languages the work is about.Language models & pretraining · 2 worksBias, fairness & safety · 1 workReasoning · 1 work

 Works by year

2020s

20251 paper20231 paper

 Papers

The most cited first (OpenAlex and our library).

  1. Catastrophic Jailbreak of Open-source LLMs via Exploiting Generation (2023 · 169 citations)
    Safety tuning of open-source LLMs is usually checked under one fixed decoding setup. For example, LLaMA2 was assessed with top-p 0.9, temperature 0.1 and a system prompt. The authors argue that this narrow testing hides serious weaknesses. They introduce the generation exploitation attack, which changes no input…
  2. On Memorization of Large Language Models in Logical Reasoning (2025 · 45 citations)
    LLMs score well on hard reasoning benchmarks yet still fail on trivial questions. This raises the suspicion that some benchmark success comes from recalling familiar problems. The authors test this idea in a controlled setting. They define LiMem, a memorization score that multiplies a model's accuracy on a puzzle set…

All works

16 works of theirs in our library, the most cited first; 2 with a page here, the others link their original until we summarise them.

  1. Detecting Pretraining Data from Large Language Models ↗arXiv:2310.16789 · no summary yetWeijia Shi et al.2023 · 224 citations
  2. Catastrophic Jailbreak of Open-source LLMs via Exploiting GenerationarXiv:2310.06987Yangsibo Huang, Samyak Gupta, Mengzhou Xia, et al.2023 · 169 citations
  3. SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal ↗arXiv:2406.14598 · no summary yetTinghao Xie et al.2024 · 87 citations
  4. Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications ↗arXiv:2402.05162 · no summary yetBoyi Wei et al.2024 · 66 citations
  5. MUSE: Machine Unlearning Six-Way Evaluation for Language Models ↗arXiv (Cornell University) · no summary yetWeijia Shi et al.2024 · 65 citations
  6. TextHide: Tackling Data Privacy in Language Understanding Tasks ↗Findings of the Association for Computational Linguistics: EMNLP 2020 · no summary yetYangsibo Huang et al.2020 · 62 citations
  7. Recovering Private Text in Federated Learning of Language Models ↗neural information processing systems · no summary yetSamyak Gupta et al.2022 · 52 citations
  8. On Memorization of Large Language Models in Logical ReasoningInternational Joint Conference on Natural Language Processing (IJCNLP)Chulin Xie, Yangsibo Huang, Chiyuan Zhang, et al.2025 · 45 citations
  9. MATH-Perturb: Benchmarking LLMs' Math Reasoning Abilities against Hard Perturbations ↗arXiv:2502.06453 · no summary yetKaixuan Huang et al.2025 · 42 citations
  10. Privacy Implications of Retrieval-Based Language Models ↗Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · no summary yetYangsibo Huang et al.2023 · 38 citations
  11. Evaluating Copyright Takedown Methods for Language Models ↗neural information processing systems · no summary yetBoyi Wei et al.2024 · 25 citations
  12. A Safe Harbor for AI Evaluation and Red Teaming ↗arXiv (Cornell University) · no summary yetShayne Longpre et al.2024 · 22 citations
  13. Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research ↗neural information processing systems · no summary yetA. Feder Cooper et al.2025 · 17 citations
  14. Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy, Research, and Practice ↗SSRN Electronic Journal · no summary yetA. Feder Cooper et al.2025 · 12 citations
  15. Fantastic Copyrighted Beasts and How (Not) to Generate Them ↗arXiv (Cornell University) · no summary yetLuxi He et al.2024 · 9 citations
  16. AI Risk Management Should Incorporate Both Safety and Security ↗arXiv (Cornell University) · no summary yetXiangyu Qi et al.2024 · 3 citations

 Co-authors here

7 co-authors with a page, the most papers together first.

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

 Cited by

Papers of our library cite their works 769 times (of the 149,439 papers whose bibliographies we read); 138 citing papers with a page here →.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/yangsibo-huang/