🧑‍🔬 Sanmi Koyejo

🏛️ Affiliations

🎓Stanford UniversityStanford · United States2014–2026 · 174 papersEmployment since 2022 · Assistant Professor, Computer Sciencecurrent🏥Stanford MedicineStanford · United States2023–2026 · 4 paperscurrent🎓Carnegie Mellon UniversityPittsburgh · United States2011–2026 · 2 paperscurrent🏥Stanford Health CareStanford · United States2024–2026 · 2 paperscurrent🎓University of Illinois Urbana-ChampaignUrbana · United States2014–2024 · 60 papersEmployment since 2015🏢GoogleMountain View · United States2019–2025 · 18 papers🎓The University of Texas at AustinAustin · United States2009–2015 · 16 papersEducation🏢Google DeepMindLondon · United Kingdom2024 · 3 papers🎓New Jersey Institute of TechnologyNewark · United StatesEducation

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 10 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.

📅 Works by year

2020s

📄 Papers

The most cited first (OpenAlex and our library).

1. Are Emergent Abilities of Large Language Models a Mirage? (2023 · 342 citations)
   The paper asks whether the "emergent abilities" reported for large language models are real. These are skills that seem to appear suddenly and unpredictably once models pass some size. The authors argue that this sharpness often comes from how performance is scored, not from abrupt changes inside the models. Using a…
2. The Leaderboard Illusion (2025 · 22 citations)
   Chatbot Arena has become the main public leaderboard for ranking LLMs through crowdsourced pairwise votes.
3. Pantograph: A Machine-to-Machine Interaction Interface for Advanced Theorem Proving, High Level Reasoning, and Data Extraction in Lean 4 (2024 · 14 citations)
   Machine learning agents that search for proofs in Lean 4 depend on an interface to the proof assistant. The standard Language Server Protocol was built for humans: it tracks cursor positions and offers no easy way to collect tactic data or to leave proof gaps for automation. Pantograph is an API and REPL written…
4. Learning to (Learn at Test Time) (2023 · 6 citations)
   Test-time training (TTT) adapts a model to each individual input through a self-supervised objective. Until now, picking an auxiliary task whose improvement actually helps the main task has depended on manual design and trial and error. Sun et al. recast supervised learning as a two-level meta-learning problem, which…

All works

27 works of theirs in our library, the most cited first; 4 with a page here, the others link their original until we summarise them.

1. Examples are not enough, learn to criticize! Criticism for Interpretability ↗neural information processing systems · no summary yetBeen Kim, Rajiv Khanna, Sanmi Koyejo2016 · 527 citations
2. Are Emergent Abilities of Large Language Models a Mirage?Conference on Neural Information Processing Systems (NeurIPS)Rylan Schaeffer, Brando Miranda, Sanmi Koyejo2023 · 342 citations
3. Rethinking machine unlearning for large language models ↗Nature Machine Intelligence · no summary yetSijia Liu et al.2025 · 197 citations
4. Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data ↗arXiv (Cornell University) · no summary yetMatthias Gerstgrasser et al.2024 · 44 citations
5. Advancing science- and evidence-based AI policy ↗Science · no summary yetRishi Bommasani et al.2025 · 25 citations
6. The Leaderboard IllusionConference on Neural Information Processing Systems (NeurIPS)Shivalika Singh, Yiyang Nan, Alex Wang, et al.2025 · 22 citations
7. 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
8. Pantograph: A Machine-to-Machine Interaction Interface for Advanced Theorem Proving, High Level Reasoning, and Data Extraction in Lean 4Lecture notes in computer scienceLeni Aniva, Chuyue Sun, Brando Miranda, et al.2024 · 14 citations
9. 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
10. End-to-End Test-Time Training for Long Context ↗arXiv (Cornell University) · no summary yetArnuv Tandon et al.2025 · 10 citations
11. Extracting books from production language models ↗arXiv (Cornell University) · no summary yetAhmed Ahmed et al.2026 · 8 citations
12. UQ: Assessing Language Models on Unsolved Questions ↗arXiv (Cornell University) · no summary yetNie, Fan et al.2025 · 6 citations
13. Learning to (Learn at Test Time)arXiv:2310.13807Yu Sun, Xinhao Li, Karan Dalal, et al.2023 · 6 citations
14. SWE-chat: Coding Agent Interactions From Real Users in the Wild ↗arXiv (Cornell University) · no summary yetJoachim Baumann et al.2026 · 5 citations
15. SpecEval: Evaluating Model Adherence to Behavior Specifications ↗arXiv (Cornell University) · no summary yetAhmed Abdelmoamen Ahmed et al.2025 · 3 citations
16. Language Models May Verbatim Complete Text They Were Not Explicitly Trained On ↗arXiv (Cornell University) · no summary yetKen Ziyu Liu et al.2025 · 3 citations
17. Reliable and Efficient Amortized Model-based Evaluation ↗arXiv (Cornell University) · no summary yetSang Truong et al.2025 · 3 citations
18. Shaping AI's Impact on Billions of Lives ↗arXiv (Cornell University) · no summary yetMariano-Florentino Cuéllar et al.2024 · 2 citations
19. Inference-Time Control for Trustworthy Large Language Models ↗Preprints.org · no summary yetYuyang Bai et al.2026
20. DecodingTrust-Agent Platform (DTap): A Controllable and Interactive Red-Teaming Platform for AI Agents ↗arXiv (Cornell University) · no summary yetZhaorun Chen et al.2026
21. Verbalizing Subliminal Learning Effects Using Text Optimization ↗arXiv (Cornell University) · no summary yetNathan Hu, Sanmi Koyejo, Christopher E. Potts2026
22. Verifiable, Articulable, and Tacit Components of Preference ↗arXiv (Cornell University) · no summary yetAlexander Spangher et al.2026
23. Neural Nonmyopic Bayesian Optimization in Dynamic Cost Settings ↗arXiv (Cornell University) · no summary yetSang Truong et al.2026
24. The Illusion of Robustness: Aggregate Accuracy Hides Prediction Flips under Task-Irrelevant Context ↗arXiv (Cornell University) · no summary yetYanzhe Zhang, Sanmi Koyejo, Diyi Yang2026
25. Structured Prompts Improve Evaluation of Language Models ↗arXiv (Cornell University) · no summary yetAsad Aali et al.2025

See all 27 works by Sanmi Koyejo →

🤝 Co-authors here

12 of 20 co-authors with a page, the most papers together first.

• Brando Miranda ×2
• Ahmet Üstün
• Alex Wang
• Beyza Ermis
• Carlos Guestrin
• Chuyue Sun
• Clark Barrett
• Daniel D'Souza
• Marzieh Fadaee
• Noah A. Smith
• Rylan Schaeffer
• Sayash Kapoor

Venues

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

• Conference on Neural Information Processing Systems (NeurIPS) ×3: NeurIPS 2025, NeurIPS 2023
• neural information processing systems ×2
• Lecture notes in computer science: Lecture notes in computer science 2024
• Nature Machine Intelligence
• Preprints.org
• SSRN Electronic Journal
• Science
• Stanford Digital Repository

🔗 Cited by

Papers of our library cite their works 917 times (of the 149,439 papers whose bibliographies we read); 247 citing papers with a page here →.

Updated 3 November 2026

↑ back to top

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