We make it easy to explore the dataset. To get started
from imagenet_x import load_annotations
table = load_annotations()
We show sample annotation responses from ImageNet. The class label is in bold, factor labels in the middle, and the free-form one-word summaries are at the bottom.

Interactive dashboard illustrating model robustness strengths and weaknesses for each ImageNet metaclass.

Robustness metaclass dashboard for Vision Transformer (ViT) ↗