A multiscale analysis of mean-field transformers in the moderate interaction regime

NeurIPSOral2025

Authors
Giuseppe Bruno, Federico Pasqualotto, Andrea Agazzi
Affiliation
Universität Bern
Venue
NeurIPS 2025
Track
Oral

TL;DR

In this paper, we study the evolution of tokens through the depth of encoder-only transformer models at inference time by modeling them as a system of particles interacting in a mean-field way and studying the corresponding dynamics. More specifically, we consider this problem in the moderate interaction regime, where the number $N$ of tokens is la…

Opening excerpt from the authors’ abstract. source

Read the paper

Topics

transformer

← All NeurIPS 2025 Oral papers · Browse the whole archive