Back to search
PyTorch implementation for Vision Transformer[Dosovitskiy, A.(ICLR'21)] modified to obtain over 90% accuracy FROM SCRATCH on CIFAR-10 with small number of parameters (= 6.3M, originally ViT-B has 86M).
Stars
203
Forks
33
Watchers
203
Open Issues
2
Overall repository health assessment
No package.json found
This might not be a Node.js project
32
commits