| Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback Y Bai, A Jones, K Ndousse, A Askell, A Chen, N DasSarma, D Drain, ... arXiv preprint arXiv:2204.05862, 2022 | 3443 | 2022 |
| A Mathematical Framework for Transformer Circuits N Elhage, 𝗡 𝗡𝗮𝗻𝗱𝗮, C Olsson, T Henighan, N Joseph, B Mann, A Askell, ... Transformer Circuits Thread, 2021 | 1216* | 2021 |
| In-context Learning and Induction Heads C Olsson, N Elhage, 𝗡 𝗡𝗮𝗻𝗱𝗮, N Joseph, N DasSarma, T Henighan, ... Transformer Circuits Thread, 2022 | 1202* | 2022 |
| Progress Measures For Grokking Via Mechanistic Interpretability 𝗡 𝗡𝗮𝗻𝗱𝗮, L Chan, T Liberum, J Smith, J Steinhardt ICLR 2023 Spotlight, 2023 | 756* | 2023 |
| Predictability and surprise in large generative models D Ganguli, D Hernandez, L Lovitt, A Askell, Y Bai, A Chen, T Conerly, ... Proceedings of the 2022 ACM Conference on Fairness, Accountability, and …, 2022 | 464 | 2022 |
| Refusal in Language Models Is Mediated by a Single Direction A Arditi, O Obeso, A Syed, D Paleka, N Rimsky, W Gurnee, 𝗡 𝗡𝗮𝗻𝗱𝗮 NeurIPS 2024, 2024 | 431* | 2024 |
| Finding Neurons in a Haystack: Case Studies with Sparse Probing W Gurnee, 𝗡 𝗡𝗮𝗻𝗱𝗮, M Pauly, K Harvey, D Troitskii, D Bertsimas Transactions on Machine Learning Research, 2023 | 299* | 2023 |
| Emergent Linear Representations in World Models of Self-Supervised Sequence Models 𝗡 𝗡𝗮𝗻𝗱𝗮, A Lee, M Wattenberg BlackboxNLP at EMNLP 2023, Honourable Mention for Best Paper, 2023 | 289* | 2023 |
| Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2 T Lieberum, S Rajamanoharan, A Conmy, L Smith, N Sonnerat, V Varma, ... Oral at BlackboxNLP 2024, 2024 | 271 | 2024 |
| Towards Best Practices of Activation Patching in Language Models: Metrics and Methods F Zhang, 𝗡 𝗡𝗮𝗻𝗱𝗮 ICLR 2024, 2023 | 172 | 2023 |
| Open Problems in Mechanistic Interpretability L Sharkey, B Chughtai, J Batson, J Lindsey, J Wu, L Bushnaq, ... arXiv preprint arXiv:2501.16496, 2025 | 169* | 2025 |
| TransformerLens: A Library for Mechanistic Interpretability of Language Models 𝗡 𝗡𝗮𝗻𝗱𝗮, J Bloom https://github.com/neelnanda-io/TransformerLens, 2023 | 159* | 2023 |
| Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoders S Rajamanoharan, T Lieberum, N Sonnerat, A Conmy, V Varma, J Kramár, ... arXiv preprint arXiv:2407.14435, 2024 | 156 | 2024 |
| A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operations B Chughtai, L Chan, 𝗡 𝗡𝗮𝗻𝗱𝗮 ICML 2023, 2023 | 148* | 2023 |
| Improving Dictionary Learning with Gated Sparse Autoencoders S Rajamanoharan, A Conmy, L Smith, T Lieberum, V Varma, J Kramár, ... NeurIPS 2024, 2024 | 146* | 2024 |
| Linear Representations of Sentiment in Large Language Models C Tigges, OJ Hollinsworth, A Geiger, 𝗡 𝗡𝗮𝗻𝗱𝗮 BlackboxNLP 2024, 2023 | 141* | 2023 |
| Does Circuit Analysis Interpretability Scale? Evidence from Multiple Choice Capabilities in Chinchilla T Lieberum, M Rahtz, J Kramár, 𝗡 𝗡𝗮𝗻𝗱𝗮, G Irving, R Shah, V Mikulik arXiv preprint arXiv:2307.09458, 2023 | 132 | 2023 |
| Transcoders Find Interpretable LLM Feature Circuits J Dunefsky, P Chlenski, 𝗡 𝗡𝗮𝗻𝗱𝗮 NeurIPS 2024, 2024 | 106* | 2024 |
| Attribution Patching: Activation Patching At Industrial Scale 𝗡 𝗡𝗮𝗻𝗱𝗮 https://www.neelnanda.io/mechanistic-interpretability/attribution-patching, 2023 | 106* | 2023 |
| Chain-of-Thought Reasoning In The Wild Is Not Always Faithful I Arcuschin, J Janiak, R Krzyzanowski, S Rajamanoharan, 𝗡 𝗡𝗮𝗻𝗱𝗮, ... arXiv preprint arXiv:2503.08679, 2025 | 98* | 2025 |
Arthur ConmyGoogle DeepMindVerified email at google.com
