#vit
- 4 Dec 2025 · Study
Transformers · Vision Transformer (ViT)
ViT treats an image as a sequence of 16×16 patches with a [CLS] token and position embeddings; patch arithmetic, and a pretrained ViT classifying a photo.
ViT treats an image as a sequence of 16×16 patches with a [CLS] token and position embeddings; patch arithmetic, and a pretrained ViT classifying a photo.