POA: Pre-training Once for Models of All Sizes

TL;DR

POA introduces a novel self-distillation framework that enables pre-training a single model to produce multiple models of different sizes simultaneously, improving efficiency and versatility for vision tasks.

Contribution

The paper proposes a tri-branch self-supervised training method with an elastic student, allowing one pre-training session to generate diverse model sizes for various downstream applications.

Findings

Achieves state-of-the-art performance on multiple benchmarks.

Produces around a hundred models of different sizes from a single pre-training.

Effective across various backbones like ViT, Swin Transformer, and ResNet.

Abstract

Large-scale self-supervised pre-training has paved the way for one foundation model to handle many different vision tasks. Most pre-training methodologies train a single model of a certain size at one time. Nevertheless, various computation or storage constraints in real-world scenarios require substantial efforts to develop a series of models with different sizes to deploy. Thus, in this study, we propose a novel tri-branch self-supervised training framework, termed as POA (Pre-training Once for All), to tackle this aforementioned issue. Our approach introduces an innovative elastic student branch into a modern self-distillation paradigm. At each pre-training step, we randomly sample a sub-network from the original student to form the elastic student and train all branches in a self-distilling fashion. Once pre-trained, POA allows the extraction of pre-trained models of diverse sizes for downstream tasks. Remarkably, the elastic student facilitates the simultaneous pre-training of multiple models with different sizes, which also acts as an additional ensemble of models of various sizes to enhance representation learning. Extensive experiments, including k-nearest neighbors, linear probing evaluation and assessments on multiple downstream tasks demonstrate the effectiveness and advantages of our POA. It achieves state-of-the-art performance using ViT, Swin Transformer and ResNet backbones, producing around a hundred models with different sizes through a single pre-training session. The code is available at: https://github.com/Qichuzyy/POA.