Going deeper with Image Transformers

TL;DR

This paper develops deeper image transformer models, optimizing their architecture and training to achieve state-of-the-art accuracy on ImageNet without external data, surpassing previous CNN and transformer models.

Contribution

It introduces two architectural modifications that enable training deeper transformers effectively, leading to improved accuracy and efficiency on image classification benchmarks.

Findings

Achieved 86.5% top-1 accuracy on ImageNet without external data.

Established new state-of-the-art on ImageNet with reassessed labels.

Produced models with less FLOPs and parameters than previous methods.

Abstract

Transformers have been recently adapted for large scale image classification, achieving high scores shaking up the long supremacy of convolutional neural networks. However the optimization of image transformers has been little studied so far. In this work, we build and optimize deeper transformer networks for image classification. In particular, we investigate the interplay of architecture and optimization of such dedicated transformers. We make two transformers architecture changes that significantly improve the accuracy of deep transformers. This leads us to produce models whose performance does not saturate early with more depth, for instance we obtain 86.5% top-1 accuracy on Imagenet when training with no external data, we thus attain the current SOTA with less FLOPs and parameters. Moreover, our best model establishes the new state of the art on Imagenet with Reassessed labels and Imagenet-V2 / match frequency, in the setting with no additional training data. We share our code and models.