Semantic Structure of Feature Space in Large Language Models

Abstract

We show that the geometric relations between semantic features in large language models' hidden states closely mirror human psychological associations. We construct feature vectors corresponding to 360 words and project them on 32 semantic axes (e.g. $\textit{beautiful-ugly}$, $\textit{soft-hard}$), and find that these projections correlate highly with human ratings of those words on the respective semantic scales. Second, we find that the cosine similarities between semantic axes are highly predictive of the correlations between these scales in the survey. Third, we show that substantial variance across the 32 semantic axes lies on a low-dimensional subspace, reproducing patterns typical of human semantic associations. Finally, we demonstrate that this geometric alignment has practical consequences for feature steering; steering a word on one semantic axis affects how the model rates that word on other semantic scales proportional to the cosine similarity between those semantic axes. These findings suggest that features should be understood through their geometric relations and the meaningful subspaces they form.