December 06, 2021
Vision transformer (ViT) models exhibit substandard optimizability. In particular, they are sensitive to the choice of optimizer (AdamW vs. SGD), optimizer hyperparameters, and training schedule length. In comparison, modern convolutional neural networks are easier to optimize. Why is this the case? In this work, we conjecture that the issue lies with the patchify stem of ViT models, which is implemented by a stride-p p×p convolution (p = 16 by default) applied to the input image. This large-kernel plus large-stride convolution runs counter to typical design choices of convolutional layers in neural networks. To test whether this atypical design choice causes an issue, we analyze the optimization behavior of ViT models with their original patchify stem versus a simple counterpart where we replace the ViT stem by a small number of stacked stride-two 3×3 convolutions. While the vast majority of computation in the two ViT designs is identical, we find that this small change in early visual processing results in markedly different training behavior in terms of the sensitivity to optimization settings as well as the final model accuracy. Using a convolutional stem in ViT dramatically increases optimization stability and also improves peak performance (by ∼1-2% top-1 accuracy on ImageNet-1k), while maintaining flops and runtime. The improvement can be observed across the wide spectrum of model complexities (from 1G to 36G flops) and dataset scales (from ImageNet-1k to ImageNet-21k). These findings lead us to recommend using a standard, lightweight convolutional stem for ViT models in this regime as a more robust architectural choice compared to the original ViT model design.
Publisher
NeurIPS
Research Topics
May 06, 2024
Haoyue Tang, Tian Xie
May 06, 2024
April 23, 2024
Jamie Tolan, Eric Yang, Ben Nosarzewski, Guillaume Couairon, Huy Vo, John Brandt, Justine Spore, Sayantan Majumdar, Daniel Haziza, Janaki Vamaraju, Theo Moutakanni, Piotr Bojanowski, Tracy Johns, Brian White, Tobias Tiecke, Camille Couprie, Edward Saenz
April 23, 2024
April 23, 2024
Sachit Menon, Ishan Misra, Rohit Girdhar
April 23, 2024
April 18, 2024
Jonas Kohler, Albert Pumarola, Edgar Schoenfeld, Artsiom Sanakoyeu, Roshan Sumbaly, Peter Vajda, Ali Thabet
April 18, 2024
Product experiences
Foundational models
Product experiences
Latest news
Foundational models