{
  "model_type": "vit",
  "architectures": [
    "ViTForImageClassification"
  ],
  "hidden_size": 768,
  "num_hidden_layers": 12,
  "num_attention_heads": 12
}