Understanding sparse autoencoder scaling in the presence of feature manifolds

2025-09-02 · Eric J. Michaud, Liv Gorton, Tom McGrath arxiv

Sparse autoencoders (SAEs) model the activations of a neural network as linear combinations of sparsely occurring directions of variation (latents). The ability of SAEs to reconstruct activations follows scaling laws w.r.t. the number of latents. In this work, we adapt a capacity-allocation model from the neural scaling literature (Brill, 2024) to understand SAE scaling, and in particular, to understand how "feature manifolds" (multi-dimensional features) influence scaling behavior. Consistent with prior work, the model recovers distinct scaling regimes. Notably, in one regime, feature manifolds have the pathological effect of causing SAEs to learn far fewer features in data than there are latents in the SAE. We provide some preliminary discussion on whether or not SAEs are in this pathological regime in the wild.

📄 PDF Abstract BibTeX arXiv:2509.02565

Code (0)

등록된 구현이 없습니다.

Scaling and evaluating sparse autoencoders

2024-06-06 · Leo Gao, Tom Dupré La Tour, Henk Tillman, Gabriel Goh 외

Sparse autoencoders provide a promising unsupervised approach for extracting interpretable features from a language model by reconstructing activations from a sparse bottleneck layer. Since language models learn many con…

Language Modelling

SAVE: Sparse Autoencoder-Driven Visual Information Enhancement for Mitigating Object Hallucination

2025-12-08 · Sangha Park, Seungryong Yoo, Jisoo Mok, Sungroh Yoon arxiv

Although Multimodal Large Language Models (MLLMs) have advanced substantially, they remain vulnerable to object hallucination caused by language priors and visual information loss. To address this, we propose SAVE (Spars…

Robust Autoencoders for Collective Corruption Removal

2023-03-06 · Taihui Li, Hengkang Wang, Peng Le, XianE Tang 외

Robust PCA is a standard tool for learning a linear subspace in the presence of sparse corruption or rare outliers. What about robustly learning manifolds that are more realistic models for natural data, such as images? …

Sparse Autoencoders Can Capture Language-Specific Concepts Across Diverse Languages

2025-07-15 · Lyzander Marciano Andrylie, Inaya Rahmanisa, Mahardika Krisna Ihsani, Alfan Farizki Wicaksono 외 arxiv

Understanding the multilingual mechanisms of large language models (LLMs) provides insight into how they process different languages, yet this remains challenging. Existing studies often focus on individual neurons, but …

Language Identification

Scaling sparse feature circuit finding for in-context learning

2025-04-18 · Dmitrii Kharlapenko, Stepan Shabalin, Fazl Barez, Arthur Conmy 외

Sparse autoencoders (SAEs) are a popular tool for interpreting large language model activations, but their utility in addressing open questions in interpretability remains unclear. In this work, we demonstrate their effe…

In-Context Learning Large Language Model