Models / timm / ViT-Base (Patch16, 224px, AugReg v2, IN21k → IN1k)

ViT-Base (Patch16, 224px, AugReg v2, IN21k → IN1k)

JSON →
vision

A Vision Transformer base model with patch size 16, pretrained on ImageNet-21k with AugReg v2 and fine-tuned on ImageNet-1k for classification.

imagevision
Specs
Context & limits
context window
max output
Lifecycle
Dates
released2021-10-01
Resources