ViT-Base (Patch16, 224px, AugReg, IN21k)

JSON →
timm vision
image

A Vision Transformer base model with patch size 16, pretrained on ImageNet-21k with AugReg for general-purpose image representation.

vision
releasedOct 2021