A Model Zoo of Vision Transformers

Damian Falk, Léo Meynent, Florence Pfammatter, Konstantin Schürholt, Damian Borth · 2025

Read A Model Zoo of Vision Transformers by Damian Falk, Léo Meynent, Florence Pfammatter, Konstantin Schürholt, Damian Borth (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-10231.txt.

Description

Damian Falk and co-authors release in this 2025 paper the first model zoo of vision transformers, the image networks now leading the field. A model zoo is a structured herd of trained networks used to study weights, compare behavior, and breed new models. Older zoos held only small, dated designs.

Their blueprint covers both pretraining and fine-tuning, and the herd holds 250 distinct models spanning wide design choices. The team checks the herd's variety with weight-space and behavior measures, then sketches uses from model analysis to generating fresh parameters. The public release lets population-style methods, once confined to toy models, run on modern designs at last. Weights and training code ship with it.

Catalog record

CREATORDamian Falk, Léo Meynent, Florence Pfammatter, Konstantin Schürholt, Damian Borth
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-10231

Browse the full library · About the Tower of Babel