A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders

TL;DR

Sparse Autoencoders (SAEs) aim to decompose the activation space of large language models (LLMs) into human-interpretable latent directions or features. As we increase the number of features in the SAE, hierarchical features tend to split into finer features (“math” may split into “algebra”, “geometry”, etc.), a phenomenon referred to as feature sp…

Opening excerpt from the authors’ abstract. source