Scaling Vision Transformers
Xiaohua Zhai, Alexander Kolesnikov, Neil Houlsby, Lucas Beyer
TLDR
This paper studies how Vision Transformers scale with model size and data, improving their architecture and training to achieve state-of-the-art ImageNet accuracy with a 2-billion parameter model.
Key contributions
- Characterized scaling laws for Vision Transformers relating error rate, data size, and compute.
- Refined ViT architecture and training to reduce memory usage and boost accuracy.
- Trained a 2-billion parameter ViT achieving 90.45% top-1 accuracy on ImageNet and strong few-shot transfer performance.
Why it matters
Understanding how Vision Transformers scale is crucial for designing more powerful and efficient models in computer vision. By establishing scaling relationships and improving training methods, this work enables the development of much larger ViT models that set new performance standards and generalize well with limited data, advancing the state of the art in visual recognition.
Original Abstract
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.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.