arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.04560cs.CVcs.AIcs.LG
收藏

缩放视觉Transformer

Scaling Vision Transformers

Google Research(谷歌研究院)

Xiaohua Zhai, Alexander Kolesnikov, Neil Houlsby, Lucas Beyer

摘要:

基于注意力的神经网络,如视觉Transformer(ViT),最近在许多计算机视觉基准测试中获得了最先进的结果。规模是获得出色结果的主要因素,因此,了解模型的缩放特性是有效设计后代的关键。虽然已经研究了缩放Transformer语言模型的规则,但还不知道视觉Transformer如何缩放。为了解决这个问题,我们缩放ViT模型和数据,向上和向下,并表征出错率,数据和计算之间的关系。在此过程中,我们改进了ViT的架构和训练,减少了内存消耗,提高了最终模型的准确性。因此,我们成功地训练了一个具有20亿个参数的ViT模型,该模型在ImageNet上达到了90.45%的top-1准确率。该模型在Few-Shot传输方面也表现良好,例如,在ImageNet上达到84.86%的top-1准确率,每个类只有10个示例。

英文摘要:

Attention-based neural networks such as the Vision Transformer (ViT) have recently attained state-of-the-art results on many computer vision benchmarks. Scale is a primary ingredient in attaining excellent results, therefore, understanding a model's scaling properties is a key to designing future generations effectively. While the laws for scaling Transformer language models have been studied, it is unknown how Vision Transformers scale. To address this, we scale ViT models and data, both up and down, and characterize the relationships between error rate, data, and compute. Along the way, we refine the architecture and training of ViT, reducing memory consumption and increasing accuracy of the resulting models. As a result, we successfully train a ViT model with two billion parameters, which attains a new state-of-the-art on ImageNet of 90.45% top-1 accuracy. The model also performs well for few-shot transfer, for example, reaching 84.86% top-1 accuracy on ImageNet with only 10 examples per class.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑