Saved in:
Bibliographic Details
Main Authors: Henry, Maxim, Deliège, Adrien, Piérard, Sébastien, Van Droogenbroeck, Marc
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.23656
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910248162295808
author Henry, Maxim
Deliège, Adrien
Piérard, Sébastien
Van Droogenbroeck, Marc
author_facet Henry, Maxim
Deliège, Adrien
Piérard, Sébastien
Van Droogenbroeck, Marc
contents Training high-capacity vision models from scratch requires substantial computational resources. To improve training efficiency of a wide target model, existing growth methods often assume the availability of narrower models, obscuring the true computational cost of the entire pipeline. We propose an efficient training protocol, RBDC, that builds wide models by coupling in a parameter-free block-diagonal way narrower, independently trained models in a recursive way. This allows a flexible allocation of the training budget available across all the models involved. Evaluated with vision transformers (DeiT) and convolutional networks (ResNet) on ImageNet, our RBDC training protocol shows a much better efficiency than models trained from scratch with the standard protocol, yielding 30% FLOPs reduction at similar test accuracies. It also achieves higher performances at same training FLOPs than training protocols from the model growth literature. Finally, we show that our models can serve as better backbones than their original counterparts for downstream object detection and instance segmentation tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2605_23656
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Recursive Block-Diagonal Coupling for Resource-Efficient Training of Vision Models
Henry, Maxim
Deliège, Adrien
Piérard, Sébastien
Van Droogenbroeck, Marc
Computer Vision and Pattern Recognition
Training high-capacity vision models from scratch requires substantial computational resources. To improve training efficiency of a wide target model, existing growth methods often assume the availability of narrower models, obscuring the true computational cost of the entire pipeline. We propose an efficient training protocol, RBDC, that builds wide models by coupling in a parameter-free block-diagonal way narrower, independently trained models in a recursive way. This allows a flexible allocation of the training budget available across all the models involved. Evaluated with vision transformers (DeiT) and convolutional networks (ResNet) on ImageNet, our RBDC training protocol shows a much better efficiency than models trained from scratch with the standard protocol, yielding 30% FLOPs reduction at similar test accuracies. It also achieves higher performances at same training FLOPs than training protocols from the model growth literature. Finally, we show that our models can serve as better backbones than their original counterparts for downstream object detection and instance segmentation tasks.
title Recursive Block-Diagonal Coupling for Resource-Efficient Training of Vision Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.23656