ParameterNet: Parameters Are All You Need

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Han, Kai, Wang, Yunhe, Guo, Jianyuan, Wu, Enhua
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916090711375872
author Han, Kai
Wang, Yunhe
Guo, Jianyuan
Wu, Enhua
author_facet Han, Kai
Wang, Yunhe
Guo, Jianyuan
Wu, Enhua
contents The large-scale visual pretraining has significantly improve the performance of large vision models. However, we observe the \emph{low FLOPs pitfall} that the existing low-FLOPs models cannot benefit from large-scale pretraining. In this paper, we introduce a novel design principle, termed ParameterNet, aimed at augmenting the number of parameters in large-scale visual pretraining models while minimizing the increase in FLOPs. We leverage dynamic convolutions to incorporate additional parameters into the networks with only a marginal rise in FLOPs. The ParameterNet approach allows low-FLOPs networks to take advantage of large-scale visual pretraining. Furthermore, we extend the ParameterNet concept to the language domain to enhance inference results while preserving inference speed. Experiments on the large-scale ImageNet-22K have shown the superiority of our ParameterNet scheme. For example, ParameterNet-600M can achieve higher accuracy on ImageNet than the widely-used Swin Transformer (81.6\% \emph{vs.} 80.9\%) and has much lower FLOPs (0.6G \emph{vs.} 4.5G). In the language domain, LLaMA-1B enhanced with ParameterNet achieves 2\% higher accuracy over vanilla LLaMA. The code will be released at \url{https://parameternet.github.io/}.
format Preprint
id arxiv_https___arxiv_org_abs_2306_14525
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle ParameterNet: Parameters Are All You Need
Han, Kai
Wang, Yunhe
Guo, Jianyuan
Wu, Enhua
Computer Vision and Pattern Recognition
The large-scale visual pretraining has significantly improve the performance of large vision models. However, we observe the \emph{low FLOPs pitfall} that the existing low-FLOPs models cannot benefit from large-scale pretraining. In this paper, we introduce a novel design principle, termed ParameterNet, aimed at augmenting the number of parameters in large-scale visual pretraining models while minimizing the increase in FLOPs. We leverage dynamic convolutions to incorporate additional parameters into the networks with only a marginal rise in FLOPs. The ParameterNet approach allows low-FLOPs networks to take advantage of large-scale visual pretraining. Furthermore, we extend the ParameterNet concept to the language domain to enhance inference results while preserving inference speed. Experiments on the large-scale ImageNet-22K have shown the superiority of our ParameterNet scheme. For example, ParameterNet-600M can achieve higher accuracy on ImageNet than the widely-used Swin Transformer (81.6\% \emph{vs.} 80.9\%) and has much lower FLOPs (0.6G \emph{vs.} 4.5G). In the language domain, LLaMA-1B enhanced with ParameterNet achieves 2\% higher accuracy over vanilla LLaMA. The code will be released at \url{https://parameternet.github.io/}.
title ParameterNet: Parameters Are All You Need
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2306.14525