Enregistré dans:
Détails bibliographiques
Auteurs principaux: Lee, Andrew, Weber, Melanie, Viégas, Fernanda, Wattenberg, Martin
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2503.21073
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911057015996416
author Lee, Andrew
Weber, Melanie
Viégas, Fernanda
Wattenberg, Martin
author_facet Lee, Andrew
Weber, Melanie
Viégas, Fernanda
Wattenberg, Martin
contents Researchers have recently suggested that models share common representations. In our work, we find numerous geometric similarities across the token embeddings of large language models. First, we find ``global'' similarities: token embeddings often share similar relative orientations. Next, we characterize local geometry in two ways: (1) by using Locally Linear Embeddings, and (2) by defining a simple measure for the intrinsic dimension of each embedding. Both characterizations allow us to find local similarities across token embeddings. Additionally, our intrinsic dimension demonstrates that embeddings lie on a lower dimensional manifold, and that tokens with lower intrinsic dimensions often have semantically coherent clusters, while those with higher intrinsic dimensions do not. Based on our findings, we introduce EMB2EMB, a simple application to linearly transform steering vectors from one language model to another, despite the two models having different dimensions.
format Preprint
id arxiv_https___arxiv_org_abs_2503_21073
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Shared Global and Local Geometry of Language Model Embeddings
Lee, Andrew
Weber, Melanie
Viégas, Fernanda
Wattenberg, Martin
Computation and Language
Machine Learning
Researchers have recently suggested that models share common representations. In our work, we find numerous geometric similarities across the token embeddings of large language models. First, we find ``global'' similarities: token embeddings often share similar relative orientations. Next, we characterize local geometry in two ways: (1) by using Locally Linear Embeddings, and (2) by defining a simple measure for the intrinsic dimension of each embedding. Both characterizations allow us to find local similarities across token embeddings. Additionally, our intrinsic dimension demonstrates that embeddings lie on a lower dimensional manifold, and that tokens with lower intrinsic dimensions often have semantically coherent clusters, while those with higher intrinsic dimensions do not. Based on our findings, we introduce EMB2EMB, a simple application to linearly transform steering vectors from one language model to another, despite the two models having different dimensions.
title Shared Global and Local Geometry of Language Model Embeddings
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2503.21073