paper-with-me

Papers

Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

2026-05-28 · Adly Templeton, Tom Conerly, Jonathan Marcus, Jack Lindsey, Trenton Bricken, Brian Chen, Adam Pearce, Craig Citro, Emmanuel Ameisen, Andy Jones, Hoagy Cunningham, Nicholas L Turner, Callum McDougall, Monte MacDiarmid, Alex Tamkin, Esin Durmus, Tristan Hume, Francesco Mosconi, C. Daniel Freeman, Theodore R. Sumers, Edward Rees, Joshua Batson, Adam Jermyn, Shan Carter, Chris Olah, Tom Henighan arxiv

We demonstrate that sparse autoencoders can extract interpretable features from Claude 3 Sonnet, a production-scale language model, addressing the open question of whether dictionary learning methods scale beyond small transformers. We trained sparse autoencoders with up to 34 million features on the model's middle layer residual stream, using scaling laws to guide hyperparameter selection. The resulting features are multilingual and multimodal (generalizing to images despite text-only training), respond to both concrete instances and abstract discussions of concepts, and can be used to steer model behavior in ways consistent with their interpretations. We find features corresponding to famous entities and locations, as well as more abstract concepts like sarcasm or errors in code. We also identify features relevant to ways in which language models might cause harm--including features representing deception, power-seeking, sycophancy, and bias--and show that these causally influence model outputs when manipulated. Additionally, we conduct analyses of feature interpretability, geometry, and computational function. However, significant limitations remain: our suite of features is incomplete, and we lack rigorous methods for evaluating whether our features faithfully capture model computations.

📄 PDF Abstract BibTeX arXiv:2605.29358

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Extracting Interaction-Aware Monosemantic Concepts in Recommender Systems

2025-11-22 · Dor Arviv, Yehonatan Elisha, Oren Barkan, Noam Koenigstein arxiv

We present a method for extracting \emph{monosemantic} neurons, defined as latent dimensions that align with coherent and interpretable concepts, from user and item embeddings in recommender systems. Our approach employs…

Beyond Interpretability: The Gains of Feature Monosemanticity on Model Robustness

2024-10-27 · Qi Zhang, Yifei Wang, Jingyi Cui, Xiang Pan 외

Deep learning models often suffer from a lack of interpretability due to polysemanticity, where individual neurons are activated by multiple unrelated semantics, resulting in unclear attributions of model behavior. Recen…

Domain GeneralizationFew-Shot Learning

MonoLoss: A Training Objective for Interpretable Monosemantic Representations

2026-02-12 · Ali Nasiri-Sarvi, Anh Tien Nguyen, Hassan Rivaz, Dimitris Samaras 외 arxiv

Sparse autoencoders (SAEs) decompose polysemantic neural representations, where neurons respond to multiple unrelated concepts, into monosemantic features that capture single, interpretable concepts. However, standard tr…

Binary Sparse Coding for Interpretability

2025-09-29 · Lucia Quirke, Stepan Shabalin, Nora Belrose arxiv

Sparse autoencoders (SAEs) are used to decompose neural network activations into sparsely activating features, but many SAE features are only interpretable at high activation strengths. To address this issue we propose t…

Encourage or Inhibit Monosemanticity? Revisit Monosemanticity from a Feature Decorrelation Perspective

2024-06-25 · Hanqi Yan, Yanzheng Xiang, Guangyi Chen, Yifei Wang 외

To better interpret the intrinsic mechanism of large language models (LLMs), recent studies focus on monosemanticity on its basic units. A monosemantic neuron is dedicated to a single and specific concept, which forms a …

DiversityFeature Correlation