| Language models represent space and time W Gurnee, M Tegmark ICLR 2024, 2023 | 436 | 2023 |
| Refusal in language models is mediated by a single direction A Arditi, O Obeso, A Syed, D Paleka, N Panickssery, W Gurnee, N Nanda Advances in Neural Information Processing Systems 37, 136037-136083, 2024 | 420* | 2024 |
| Finding Neurons in a Haystack: Case Studies with Sparse Probing W Gurnee, N Nanda, M Pauly, K Harvey, D Troitskii, D Bertsimas Transactions of Machine Learning Research (TMLR), 2023 | 299* | 2023 |
| Emergent introspective awareness in large language models J Lindsey arXiv preprint arXiv:2601.01828, 2026 | 230* | 2026 |
| Not All Language Model Features Are One-Dimensionally Linear J Engels, EJ Michaud, I Liao, W Gurnee, M Tegmark The Thirteenth International Conference on Learning Representations, 2024 | 131* | 2024 |
| Circuit tracing: Revealing computational graphs in language models E Ameisen, J Lindsey, A Pearce, W Gurnee, NL Turner, B Chen, C Citro, ... Transformer Circuits Thread 6, 2025 | 130 | 2025 |
| Learning sparse nonlinear dynamics via mixed-integer optimization D Bertsimas, W Gurnee Nonlinear Dynamics 111 (7), 6585-6604, 2023 | 76 | 2023 |
| Universal neurons in GPT2 language models W Gurnee, T Horsley, ZC Guo, TR Kheirkhah, Q Sun, W Hathaway, ... Transactions of Machine Learning Research (TMLR), 2024 | 74* | 2024 |
| The remarkable robustness of llms: Stages of inference? V Lad, JH Lee, W Gurnee, M Tegmark arXiv preprint arXiv:2406.19384, 2024 | 71* | 2024 |
| Fairmandering: A column generation heuristic for fairness-optimized political districting W Gurnee, DB Shmoys SIAM Conference on Applied and Computational Discrete Algorithms (ACDA21), 88-99, 2021 | 56 | 2021 |
| Confidence regulation neurons in language models A Stolfo, B Wu, W Gurnee, Y Belinkov, X Song, M Sachan, N Nanda Advances in Neural Information Processing Systems 37, 125019-125049, 2024 | 36* | 2024 |
| Combatting gerrymandering with social choice: The design of multi-member districts N Garg, W Gurnee, D Rothschild, D Shmoys Proceedings of the 23rd ACM Conference on Economics and Computation, 560-561, 2022 | 15 | 2022 |
| Sae reconstruction errors are (empirically) pathological W Gurnee AI Alignment Forum, 16, 2024 | 12* | 2024 |
| Multilevel interpretability of artificial neural networks: leveraging framework and methods from neuroscience Z He, J Achterberg, K Collins, K Nejad, D Akarca, Y Yang, W Gurnee, ... arXiv preprint arXiv:2408.12664, 2024 | 10 | 2024 |
| Training Dynamics of Contextual N-Grams in Language Models L Quirke, L Heindrich, W Gurnee, N Nanda NeurIPS 2023 Workshop on Attributing Model Behavior at Scale, 2023 | 9 | 2023 |
| Tracing attention computation: Attention connects features, and features direct attention H Kamath, E Ameisen, I Kauvar, R Luger, W Gurnee, A Pearce, ... Transformer Circuits Thread, 2025 | 7 | 2025 |
| When models manipulate manifolds: The geometry of a counting task W Gurnee, E Ameisen, I Kauvar, J Tarng, A Pearce, C Olah, J Batson arXiv preprint arXiv:2601.04480, 2026 | 3 | 2026 |
| Scalable approximations of capacitated k-medians for political districting W Gurnee Technical report, Cornell University, Ithaca, United States, 2020 | 1 | 2020 |
| Towards an Artificial Neuroscience: Analytics for Language Model Interpretability RW Gurnee Massachusetts Institute of Technology, 2025 | | 2025 |
Neel NandaMechanistic Interpretability Team Lead, Google DeepMindVerified email at deepmind.com
