Generalized Linear Mode Connectivity for Transformers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Theus, Alexander, Cabodi, Alessandro, Anagnostidis, Sotiris, Orvieto, Antonio, Singh, Sidak Pal, Boeva, Valentina
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915614290870272
author Theus, Alexander
Cabodi, Alessandro
Anagnostidis, Sotiris
Orvieto, Antonio
Singh, Sidak Pal
Boeva, Valentina
author_facet Theus, Alexander
Cabodi, Alessandro
Anagnostidis, Sotiris
Orvieto, Antonio
Singh, Sidak Pal
Boeva, Valentina
contents Understanding the geometry of neural network loss landscapes is a central question in deep learning, with implications for generalization and optimization. A striking phenomenon is linear mode connectivity (LMC), where independently trained models can be connected by low- or zero-loss paths despite appearing to lie in separate loss basins. However, this is often obscured by symmetries in parameter space -- such as neuron permutations -- which make functionally equivalent models appear dissimilar. Prior work has predominantly focused on neuron reordering through permutations, but such approaches are limited in scope and fail to capture the richer symmetries exhibited by modern architectures such as Transformers. In this work, we introduce a unified framework that captures four symmetry classes -- permutations, semi-permutations, orthogonal transformations, and general invertible maps -- broadening the set of valid reparameterizations and subsuming many previous approaches as special cases. Crucially, this generalization enables, for the first time, the discovery of low- and zero-barrier linear interpolation paths between independently trained Vision Transformers and GPT-2 models. Furthermore, our framework extends beyond pairwise alignment to multi-model and width-heterogeneous settings, enabling alignment across architectures of different sizes. These results reveal deeper structure in the loss landscape and underscore the importance of symmetry-aware analysis for understanding model space geometry.
format Preprint
id arxiv_https___arxiv_org_abs_2506_22712
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Generalized Linear Mode Connectivity for Transformers
Theus, Alexander
Cabodi, Alessandro
Anagnostidis, Sotiris
Orvieto, Antonio
Singh, Sidak Pal
Boeva, Valentina
Machine Learning
Understanding the geometry of neural network loss landscapes is a central question in deep learning, with implications for generalization and optimization. A striking phenomenon is linear mode connectivity (LMC), where independently trained models can be connected by low- or zero-loss paths despite appearing to lie in separate loss basins. However, this is often obscured by symmetries in parameter space -- such as neuron permutations -- which make functionally equivalent models appear dissimilar. Prior work has predominantly focused on neuron reordering through permutations, but such approaches are limited in scope and fail to capture the richer symmetries exhibited by modern architectures such as Transformers. In this work, we introduce a unified framework that captures four symmetry classes -- permutations, semi-permutations, orthogonal transformations, and general invertible maps -- broadening the set of valid reparameterizations and subsuming many previous approaches as special cases. Crucially, this generalization enables, for the first time, the discovery of low- and zero-barrier linear interpolation paths between independently trained Vision Transformers and GPT-2 models. Furthermore, our framework extends beyond pairwise alignment to multi-model and width-heterogeneous settings, enabling alignment across architectures of different sizes. These results reveal deeper structure in the loss landscape and underscore the importance of symmetry-aware analysis for understanding model space geometry.
title Generalized Linear Mode Connectivity for Transformers
topic Machine Learning
url https://arxiv.org/abs/2506.22712