Neel Nanda

Artificial Intelligence · 2022–2024

A researcher in the Science corpus: 8 papers; trusts 57 researchers and is trusted by 3 researchers here, as inferred from the citation graph.

Subfields

• Artificial Intelligence — 250 citations into it; #24,060 of 35,345 by trust here, #3,080 by citations
• Safety, Robustness & Interpretability — 71 citations into it
• Large Language Models — 65 citations into it
• Computer Vision and Pattern Recognition — 62 citations into it
• Natural Language Processing — 48 citations into it
• Evaluation & Benchmarks — 37 citations into it
• Representation & Self-Supervised Learning — 21 citations into it
• Cognitive Neuroscience — 7 citations into it; #9,355 of 10,852 by trust here, #8,086 by citations

Whom Neel Nanda trusts overall

Their strongest outbound trust, strongest first, inferred from their citations.

1. Alec Radford
2. Geoffrey E. Hinton
3. Yoshua Bengio
4. Ilya Sutskever
5. Alex Krizhevsky
6. Diederik P. Kingma
7. Tomas Mikolov
8. Yonatan Belinkov
9. Kaiming He
10. Kenneth Li

Who trusts Neel Nanda overall

1. Michael R Douglas
2. Leonard Bereska
3. Wes Gurnee
4. Chris Wendler
5. Efstratios Gavves
6. Shizhuo Dylan Zhang
7. Veniamin Veselovsky
8. Max Tegmark
9. Arthur Conmy

Papers from Neel Nanda's perspective

Ordered by how strongly the researchers Neel Nanda trusts endorse each paper. Same corpus, a different order for every researcher.

1. ImageNet classification with deep convolutional neural networks
2. Attention is All you Need (Neural Information Processing Systems)
3. Deep Residual Learning for Image Recognition (Computer Vision and Pattern Recognition)
4. Adam: A Method for Stochastic Optimization (International Conference on Learning Representations)
5. Long Short-Term Memory
6. Gradient-based learning applied to document recognition
7. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (North American Chapter of the Association for Computational Linguistics)
8. ImageNet: A large-scale hierarchical image database
9. Sequence to Sequence Learning with Neural Networks (Neural Information Processing Systems)
10. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift (International Conference on Machine Learning)

Their papers

• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback (2022) — 4,254 citations
• Predictability and Surprise in Large Generative Models (2022) — 365 citations
• In-context Learning and Induction Heads (2022) — 942 citations
• Progress measures for grokking via mechanistic interpretability (2023) — 912 citations
• Refusal in Language Models Is Mediated by a Single Direction (2024) — 866 citations
• Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2 (2024) — 377 citations
• Finding Neurons in a Haystack: Case Studies with Sparse Probing (2023) — 363 citations
• Emergent Linear Representations in World Models of Self-Supervised Sequence Models (2023) — 353 citations